IBM 4341 Processor Model Groups 2, 11 and 12 Channel...

Systems

GA24-3780-1 File No. 4300-01

I BM 4341 Processor Model Groups 2, 11, and 12 Channel Characteristics

--...------- ------- _.. __ - --. ---- -- --------_ _..._,_

Preface

This manual describes how to check the effects of loads imposed on the channels of the IBM 4341 Processor Model Groups 2, 11 and 12. With this information, programmers and systems analysts can verify that a proposed configuration of input/ output (1/0) devices work satisfactorily with these 4341 Processors.

This manual gives information on channel organization and implementation, subchannels and unit control words, procedures for channel interference determination, concurrent 1/0 capabilities, block-multiplexer and byte-multiplexer channel loading, specific device considerations, and channel evaluation tables.

Before using this manual, you should have a thorough understanding of input/ output operations for the 4341 as described in:

• IBM 4341 Processor Model Group 2 Functional Characteristics and Processor Complex Configurator, Order No. GA24-3763

Second Edition (October 1982)

IBM 4341 Processor Model Groups 9, JO, 11, and 12 Functional Characteristics, GA24-3797

IBM System/ 3 70 Principles of Operation, GA22-7000

IBM 4300 Processors Principles of Operation for ECPS:VSE Mode, GA22-7070.

When testing for data overrun on the byte-multiplexer channel, two special worksheets are required:

IBM 4341 Byte-Multiplexer Channel Preliminary Worksheet, GX24-3746

• IBM 4341 Processors Channel Load Sum Worksheet, GX24-3670.

This major revision obsoletes GA24-3780-0 and Technical Newsletter GN24-0932 (31 March 1982). This edition adds information for the 4341 Processor Model Group 12. Technical changes and additions to the text and illustrations are indicated by a line to the left of the change.

Information contained in this manual is subject to change from time to time. Any such change will be reported in subsequent revisions or through the System Library Subscription Service.

It is possible that this material may contain reference to, or information about, IBM products (machines and programs), programming, or services that are not announced in your country. Such references or information must not be construed to mean that IBM intends to announce such IBM products, programming, or services in your country.

Requests for copies of IBM publications should be made to your IBM representative or to the IBM branch office serving your locality.

A form for reader's comments is provided at the back of the publication. If the form has been removed, comments may be addressed to the IBM General Technology Division, Product Publications, Dept. KIO, P.O. Box 6, Endicott, NY 13760. IBM may use or distribute any of the information you supply in any way that it believes appropriate without incurring any obligation whatever.

© Copyright International Business Machines Corporation 1981, 1982

...

• '

Contents

IBM 4341 Processor Model Groups 2, 11, and 12 Channel Characteristics 5

General Channel Information 5 Channel Control 6

Channel Registers 7 Chaining 7 Fetching Channel Command Words 7 Data Chaining in Gaps 8 Late Command Chaining 9 Storage Addressing 9

Channel Implementation 9 Byte-Multiplexer Channel 9 Block-Multiplexer Channels 10 Channel-Available Interruption 10 Channel Priority 11 Function of Traps 11 Channel Data Buffer 11

Subchannels and Unit Control Words 12 Subchannels 12 Unit Control Words (UCWs) 12

UCW Pool 12 Channel UCW Directory 13 UCW Assignment 13 Device Considerations 13

Channel Interference with Processor 16 Calculation of Total Channel/Processor Interference

Time 16 Calculation of Total Channel/Processor Interference

Percentage 18 Program Overrun 19

Assessing Program Overrun 19

Concurrent Input/Output Capabilities 21 Worst-Case Loads 21

Overrun 21 Loss of Performance 21

Conventions for Satisfactory Channel Programs 21 Data Chaining Considerations 22 Relationship of Conventions and Evaluation

Procedures 22 Scope of Conventions 22 Classes of Commands 23 Command Classification for 1/0 Devices 24

Evaluating Heavily Loaded Channels 27

Block-Multiplexer Channel Loading 28 Overrun Evaluation Consideration 28 Late Command Chaining 28 Testing for Overrun 28

Byte-Multiplexer Channel Loading 29 Byte-Mode Considerations 29

Device Load 29 Device Wait Time 29 Device Priority on Byte-Multiplexer Channel 30 Interference from Priority Devices 31

Byte-Mode Evaluation Procedures 36 IBM 3704/3705 Calculations 49

Appendix A. Evaluation Factors for Channel 0 or Channel 5 (Model Group 12) 55

Table A-1. 1255 and 1287 55 Table A-2. 1288 and 1419 56 Table A-3. 1442 and 2501 Bl and B2 57 Table A-4. 2520 58 Table A-5. 2701 59 Table A-6. 7770 60

Appendix B. Evaluation Factors for Byte-Multiplexer Channel 4 61

Table B-1. 1255 and 1287 61 Table B-2. 1288 and 1419 62 Table 8-3. 1442 and 2501 Bl and B2 63 Table B-4. 2520 64 Table B-5. 2701 65 Table B-6. 7770 66

Appendix C. Device Data Rates 67 Table C-1. Data Rates by Device 6 7 Table C-2. Data Rates by Device 68

Appendix D. Miscellaneous Tables 69 Table D-1. Data Chaining Rates for Block-Multiplexer

Channels 69 Table D-2. 3704/3705 Timing Chart 70

Index 71

Contents 3

Figures

Figure 1. Sharing of Processor and Channel Facilities 8

Figure 2. UCW Assignment Chart 14

Figure 3. Processor Interference Times in Microseconds (System/370 Mode) 17

Figure 4. Processor Interference Times in Microseconds (ECPS:VSE Mode) 17

Figure 5. Duration of Channel Interference with Processor - (Example Calculation for a Tape-to-Printer Operation in System/370 Mode) 18

Figure 6. Priority Load Curve 32

Figure 7. Use of Priority Time Factors 33

Figure 8. Example of Priority Device Causing Interference by Command Chaining (A-Factor Interference) 34

Figure 9. Example of Priority Load Curve of a Priority Device Causing Interference by Transferring Data (B-Factor Interference) 34

Figure 10. Examples Showing How to Choose Priority Load Factors 35

Figure 11. Sample Byte-Multiplexer Channel Preliminary Worksheet 37

Figure 12. Sample Channel Load Sum Worksheet 38

Figure 13. Example of Preliminary Worksheet Entries 39

Figure 14. Example of Final Worksheet Entries 40

Figure 15. Example of Preliminary Worksheet with a 2501-82 and a 1288 47

Figure 16. Example of Final Worksheet with a 2501-82 and a 1288 48

4 IBM 4341 Processor Channel Characteristics

f .,

I IBM 4341 Processor Model Groups 2, 11 and 12 Channel Characteristics

This publication describes the calculations required to optimize channel operation on the IBM 4341 Processor Model Groups 2, 11, and 12. Detailed formulas, along with timing charts for input/ output devices, enable users to determine maximum channel throughput with data overrun avoidance for actual configurations.

General Channel Information I The channels of the 4341 Processor Model Groups 2, 11, and 12 transfer data

between main storage and I/ 0 devices under control of a channel program that is executed independently of the processor program. The processor is free to resume the processor program after initiating an I/ 0 operation.

I The channels can run concurrently within the data-rate and channel-programming conventions specified in this manual.

A major feature of the channels is the common 1/0 interface connection between a variety of 1/0 devices and System/360, System/370, and 4300 Processor input/ output control units.

At the end of an I/ 0 operation, the channel signals an I/ 0 interruption request to the processor. If not disallowed, an 1/0 interruption occurs that places the 1/0 new program status word (PSW) in control of the processor. When 1/0 interruptions are disallowed, interruption requests are queued (stacked). Until honored, an I/ 0 interruption condition is called a pending I I 0 interruption.

At the end of an I/ 0 interruption, a channel has information about the success or lack of success of the operation. The information (status) is available to the processor program.

Each channel can perform these functions:

• Accept an 1/0 instruction from the processor.

• Address the device specified by an I/ 0 instruction.

• Fetch the channel program from main storage.

• Decode the channel command words that make up the channel program.

• Test each channel command word (CCW) for validity.

• Execute CCW functions.

• Place control signals on the 1/0 interface.

• Accept control-response signals from the 1/0 interface.

• Transfer data between an I/ 0 device and main storage.

• Check parity of bytes transferred.

• Count the number of bytes transferred.

• Accept status information from I/ 0 devices.

• Maintain channel-status information.

• Signal interruption requests to the processor.

• Sequence interruption requests from I/ 0 devices.

• Send status information to location 40 hexadecimal when an interruption occurs.

• Send status information to location 40 hexadecimal upon processor request.

Introduction 5

Channel Control The channels provide a common input/ output interface that permits the attachment of control units. The control units are governed by six channel commands and seven processor instructions. The instructions are:

Start 1/0 Start 1/0 Fast Release Test Channel Test 1/0 Halt 1/0 Halt Device Clear 1/0.

All 1/0 instructions set the PSW condition code. Under certain conditions, all except Test Channel cause a channel status word to be stored. A Test Channel instruction calls for information about the addressed channel; a Test 1/0 instruction calls for information about a channel and a particular device. Halt 1/0 terminates any operation on the addressed channel, subchannel, or device. Halt Device terminates an 1/0 operation on a particular device on a channel without interfering with other 1/0 operations in progress on the channel. Only Start 1/0 and Start 1/0 Fast Release make .use of channel command words.

A Start 1/0 instruction initiates execution of one or more 1/0 operations, and specifies a channel, subchannel, control unit, and 1/0 device. The channel fetches the channel address word (CAW) from location 48 hexadecimal. The CAW contains the protection key and the address of the first channel command word for the operation. Beginning with the first CCW specified by the CAW, the channel fetches and executes one or more CCWs.

I Note: The 4341 Model Group 12 implements the Start I I 0 Fast Release instruction. However, in the 4341 Model Groups 2 and 1 J, the Start 1/0 Fast Release instruction is decoded and executed as a Start I I 0 instruction. Throughout this manual, references to Start I I 0 also apply to Start I I 0 Fast Release.

The Clear 1/0 instruction causes the current operation being executed by the addressed device to be terminated with both byte-multiplexer and block-multiplexer channels. In selector channel mode, the instruction is executed as a Test 1/0.

Six channel commands are used:

Read Write Read Backward Control Sense Transfer-in-Channel (TIC).

The first three are self-explanatory. Control commands specify such operations as advance paper in a printer, set tape

density, and rewind tape. A Sense command brings information from a control unit into main storage.

This information can concern unusual conditions detected during the latest 1/0 operation and detailed status about the device.

A Transfer-in-Channel (TIC) command specifies the location in main storage from which the next CCW in the channel program is to be fetched. A TIC cannot specify another TIC. Also, the CAW cannot address a TIC.

Each CCW specifies the channel operations to be performed and, for ...,, data-transfer operations, specifies contiguous locations in main storage to be used. One or more CCWs make up a channel program that directs a channel operation.


Channel Registers

Chaining

Command retry is a channel and control-unit procedure that causes a command

to be retried without requiring an I/O interruption. Retry is initiated by the control unit. When the command being executed

encounters a retriable error, the control unit presents unit check and status modifier (retry status) along with channel end, or channel end and device end, to

the channel. If conditions permit, a normal device reselection occurs to reissue the

previous command. If retry is not possible, any chaining is terminated, and an I/O

interruption follows.

For each I/0 device selected, the channels maintain the following channel-control

information:

Protection key Data address Identity of operation specified by command code CCW flags Byte count Channel status Address of next CCW

On byte- and block-multiplexer channels, the listed information must be

maintained for each subchannel in operation. Storage for this information is

provided by special channel storage that is not directly addressable. Each

subchannel has provision in a special area of storage for channel-register

information. When a particular subchannel is selected by a Start I/0 instruction

and a channel program is initiated, the channel storage locations for the

subchannel are loaded with the information necessary for each operation of the

subchannel. At the completion of each activity in a subchannel, its particular area in storage

contains updated information, and the multiplexer channel is available for

operation of another subchannel. Figure 1 shows the sharing of facilities by the channels and the processor.

A single CCW may specify contiguous locations in main storage for a data-transfer

operation, or successive CCWs may be chained together to specify a set of noncontiguous storage areas. The presence of a flag bit in a CCW causes chaining to the next CCW.

In data chaining, the address and count information in a new CCW is used; the

command code field is ignored. Entire CCWs, including their command code fields, can also be chained together

for a sequence of channel operations. Such coupling, called command chaining, is specified by a different flag bit in a CCW.

Data chaining has no affect on a device, as long as the channel has sufficient

time to perform both data chaining and data transfer for the device. In this manual, when a device is said to data-chain, the channel program for the

device specifies data chaining.

Fetching Channel Command Words The channel must fetch a new CCW whenever a CCW specifies data chaining,

command chaining, or transfer in channel (TIC). The extra control activity caused

by these operations takes time and diminishes a channel's ability to do other work.

Introduction 7

Data Chaining in Gaps

Execute Yes •-----1Channe1

Trap

Execute Yes • Required

Trap

Main Program Routine: Execute

Next Microinstruction

'--~~~~~~~.-~~~____.I No A trap is an interruption of the current activity to allow another activity of a higher priority to take precedence.

Figure 1. Sharing of Processor and Channel Facilities

A data-chaining fetch usually occurs while a channel also has a data-transfer load from the same device. The time required to fetch the new CCW necessarily limits the time available for successive data transfers through the channel. An absence of data chaining ordinarily permits a channel to operate with a faster 1/0 device. Similarly, when a channel is not transferring data, a data-chaining operation has a lesser impact on channel facilities.

For direct-access storage devices, such as an IBM 3330 Direct Access Storage or an IBM 2314 Direct Access Storage Facility, formatting Write commands causes the control unit to create gaps between count, key, and data fields on the recording track. Read, Write, and Search commands that address more than one of the fields can specify data chaining to define separate areas in main storage for the fields.

The gaps on a track have significance to channel-programming considerations for direct-access storage devices. The channel does not transfer data during the time a gap is created or passes under the read-write head. This time is sufficient for the processor to perform a command-chaining or data-chaining operation.

Command chaining ordinarily occurs only during gap time, but data chaining can occur during gap time or while data is being transferred. A data-chaining operation occurring during gap time has less impact on channel facilities than when data transfers also occur. If a channel program for a direct-access storage device


•

Late Command Chaining

Storage Addressing

calls for data chaining only during gap time, the device's overall load on channel

facilities is significantly less. When a direct-access device is said to data-chain in a gap, the reference is to a

gap other than a gap following a data field. The latter causes a device-end

indication. Command chaining is used in such a gap if the transfer or more

information is desired. Device end occurring in the absence of a CCW specifying

command terminates the operation. When command chaining continues the

operation, the status information available at the end of the operation relates to

the last operation in the chain. During a read operation, an attempt to data-chain in a gap following a data field

causes an incorrect length indication in the channel status byte.

Operation of direct-access devices, such as disk storage, requires command

chaining. Between certain operations, such as the search for a record identification

key and the reading of a data field on a direct access-storage device, the control

unit has a fixed time interval during which it must receive and execute a new

command. If activity on other channel(s) causes too much delay in initiation of

the operation specified by the new command, the channel program is terminated

and an I/0 interruption occurs. If, however, the device has command retry

capabilities, the command is retried.

During data chaining, the beginning and ending byte addresses and the minimum

number of bytes transferred are factors in the maximum data rates that different

System/360 and System/370 models and 4300 Processors channels can sustain.

When writing channel programs for larger systems, consider the storage width and

the possibility of adding faster I/ 0 devices.

Channel Implementation

Byte-Multiplexer Channel

The 4341 Processor Model Groups 2, 11, and 12 have two basic types of

channels. A byte-multiplexer channel (Channel 0) and five block-multiplexer

channels (Channels 1 - 5) are standard. On the Model Groups 2 and 11, a second

byte-multiplexer channel can be selected on channel 4. On the Model Group 12,

channel 4 or 5 may be selected as a second byte-multiplexer channel.

The channels are integrated. They share with the processor the use of control

storage, the processor storage data flow, and the arithmetic logic unit. Each

channel can attach as many as eight control units and can address up to 256 1/0 devices, except for the byte-multiplexer channel 0, which can only attach up to

seven control units because addresses OFO-OFF are reserved. Control units are

connected to all channels through the standard 1/0 interface.

The byte~multiplexer channels have a single data path that can be fully utilized by

one I/ 0 device (burst mode) or shared by many I/ 0 devices (multiplex mode).

Burst mode is defined as operation in which the device and the channel remain

connected for a relatively long period of time in terms of system operation.

Multiplex mode is defined as byte-interleaved operation in which the channel and

any one device remain connected for a relatively short period of time, typically

long enough to transfer one byte or a small number of bytes. Multiplexing,

however, refers to the channel and device capability of disconnecting and

reconnecting during an operation. The byte-multiplexer channel operates either in

burst mode or in multiplex mode with multiplexing capability between bytes or

groups of bytes.

Introduction 9

Some control units are designed to operate on the byte-multiplexer channel in burst mode or in multiplex mode. Other control units allow the mode to be set at the installation.

When multiple I/O devices concurrently share byte-multiplexer channel facilities, the operations are in multiplex mode. Each device is selected, one at a time, for transfer of a byte or a few bytes to or from main storage. Bytes from multiple devices are interleaved and routed to or from desired locations in main storage. Thus, the byte-multiplexer channel data path is used by one device for transfer of one or a few bytes of data, and then another device uses the same data path. The sharing of the data path makes each device appear to the programmer as if it has a data path of its own. A device's share of the data path is called a subchannel.

Block-Multiplexer Channels Five block-multiplexer channels are standard. Block-multiplexer channel operation is described in the IBM 4300 Processors Principles of Operation for ECPS:VSE Mode, Order No. GA22-7070, and IBM System/ 3 70 Principles of Operation, GA22-7000.

The block-multiplexer channels can operate in either block-multiplex mode or selector mode. The mode of operation is determined by the block-multiplex mode bit of the control registers, as well as how the the I/0 device is described to the system.

The block-multiplexer channel in block-multiplex mode allows concurrent operation of many 1/0 devices. This mode is intended for relatively high-speed burst operations. A single block-multiplexer channel can support interleaved, concurrent execution of multiple high-speed channel programs. The block-multiplexer channel in this mode can be shared by multiple high-speed I/0 devices operating concurrently just as the byte-multiplexer channel can be shared by multiple low-speed devices. Like the byte-multiplexer channel, the block-multiplexer channel has multiple subchannels, each has an associated UCW in control storage and can support one 1/0 operation.

A block-multiplexer channel functions differently in selector mode in the way that it handles command-chained channel programs. A block-multiplexer channel operating in selector mode and executing a command-chained channel program is busy during the entire time the channel program is in operation, whether or not data transfer is occurring. A block-multiplexer channel in block-multiplex mode executing a command-chained channel program can disconnect from the operational program during certain nondata transfer operations; that is, a block-multiplexer channel in block-multiplex mode can be freed during nonproductive activity (for example, during disk seeking and most record positioning) to allow more data to be transferred per unit of channel busy time.

The selector and block-multiplex modes monitor the interface for an expected tag sequence. If an erroneous tag is activated (by a control unit) in addition to the normal interface sequence, the channel may not indicate a check. The channel may ignore the erroneous tag and proceed as if it did not exist.

Channel-Available Interruption The processor implements the channel-available interruption on channels 1 through 5 when they are operating as block-multiplexer channels.

The channel-available-interruption condition is meaningful only for the block-multiplexer. In selector mode, the channel always terminates a channel-busy condition with an interruption or an interruption-pending condition.


Channel Priority

Function of Traps

Channel Data Buffer

All channels share one trap level for all functions, such as data transfer, chaining, status handling, and UCW fetching and storing.

The priority for handling channel trap requests is channel 1, 2, 3, 4, 5, and 0. To prevent high priority channels with constant trap requests from locking out low priority channels, each channel, except channel 0, has a trap request inhibit latch. When a channel's inhibit latch is on or when a channel is in a trap, the trap priority controls do not consider a pending trap request for that channel. When a channel trap is taken, the inhibit lateh for that channel is set on, or all inhibit latches are reset off. That is, the inhibit latch for the trapping channels is set if any trap requests are pending for lower priority channels; otherwise, all inhibit latches are reset.

Traps are requested by the channel hardware to indicate to the microcode that certain events have occurred or that certain conditions exist. Examples: the last command issued to the channel hardware has been completed, a device has connected to the 1/0 interface because of a polling sequence, or there is data in the data buffer to be transferred to storage.

The functions of the channel-trap microcode include: fetching and storing the UCWs; transferring data between storage and the channel data buffers; fetching CCWs for SIO, command chaining, and data chaining; handling status and interface check conditions detected by the hardware; etc.

When the channel operation is set up, hardware controls the data transfer. Each channel has 256 bytes of channel data buffer. Depending on the data length and address boundary, hardware controls the transfer of data to processor storage:

• in 64-byte blocks • in a partial block to line up to a 64-byte address boundary • in a partial block to complete a data transfer • in a partial block for very short records • in a partial block for byte-multiplex operation.

Introduction 11

Subchannels and Unit Control Words

Subchannels The channel facilities required to sustain a single 1/0 operation are called a subchannel. Subchannels can be either unshared or shared. An unshared subchannel has the facilities to operate only one 1/0 device. A shared subchannel provides facilities to operate one of an attached set of 1/0 devices.

The initiation of multiple 1/0 operations requires that the subchannels have channel storage to record the addresses, count, status, and control information associated with the I/ 0 operation. The storage for a single set of such information is called a unit control word (UCW). The UCW storage is provided by moving the high address boundary of user storage. The number of required UCWs is determined by the number of 1/0 devices described to the system. A maximum of 1,024 UCWs are available.

When an 1/0 operation is initiated, the UCW specified by the 1/0 unit address is modified (if the subchannel is not already busy) according to the operation to be performed. Each channel has one local-storage area for use in processing a single UCW. Therefore, when a UCW is set up in that area during execution of an 1/0 operation, requests from other units on that channel are prevented.

Unit Control Words (UCWs)

UCW Pool

A UCW is either shared or unshared according to the type of subchannel it contains. The subchannel type is determined by the control and device(s) that use the subchannel and the UCW. An unshared UCW is reserved for a specified device. Examples are:

1. A single control unit that controls one I/ 0 unit and requires one unshared UCW. For example, the 1443 Printer or the 3211 Printer each requires an unshared UCW.

2. A single unit that contains several control units and requires an unshared UCW for each unit. For example, the 2821 Control Unit that handles functions for a 1403 Printer and the reader and punch sections of a 2540 Read/Punch requires three unshared UCWs.

3. A single control unit that services the requirements of several devices at once and requires an unshared UCW for each device. For example, the 3830 Storage Control requires an unshared UCW for each device.

A shared UCW is reserved for a specified control unit and is shared by the devices attached to that control unit. For example, each 3272 Control Unit requires an exclusive UCW, but all 3277 Displays serviced by that control unit share the UCW assigned to the 3272.

From 128 to 1,024 subchannels (UCWs) can be configured. If more than 128 UCWs are required, additional groups of 32 UCWs are allocated, up to the maximum of 1,024.

Each of the minimum allocation of 128 UCWs occupies 64 bytes of processor storage (for a total of 8,192 bytes). Additional UCWs (in groups of 32, up to a maximum of 1,024) may be assigned with the user-storage reduction described in the following section.


•

System Storage Requirements

Channel UCW Directory

UCW Assignment

Device Considerations

A portion of processor storage is required for systems functions and reduces the amount of processor storage available for user programming. The size of the reduction depends upon the processor configuration.

Each group of 32 additional UCWs reduces the usable processor storage by 2,048 bytes. For example, 128 UCWs require 8K of storage, and 1,024 UCWs require 64K of storage. For each increase in the number of UCWs, the additional UCW storage requirement limits the high address boundary.

Sixteen UCWs (000 through OOF) are reserved for natively attached I/O and for internal functions. Each channel has a shared UCW through which all unassigned 1/0 devices can present asynchronous interruptions. Each device attached to a channel must have a UCW associated with its address.

Each channel has a channel directory. Each directory has 256 entries, one for each possible device address (00-FF) on the channel.

Each directory entry contains the reference number of its associated UCW. Each entry also contains the following status and attribute bits:

ASSIGNED

SHARED SEL DATA STREAM

indicates that this entry was assigned to an 1/0 device by the user. indicates a shared UCW. indicates a device that must operate in selector mode. indicates a control unit or device attached to a control unit that can operate in data streaming mode.

UCWs for natively attached 1/0 devices (such as the 3278-2A) are preassigned. The logical addresses for these devices are assigned by the user. All other I/ 0 devices must be described to the processor.

The device addresses for a channel that are not used by a described device are assigned to a channel shared UCW. A device that has not been described to the sy~tem can submit status via the channel shared UCW, but 1/0 instructions issued to that device result in a not operational condition (condition code 3). This information is kept in the UCW directory tables for automatic assignment of UCWs by the processor.

All UCW assignments are written onto the system diskette and become effective after subsequent initial microcode load (IML) operations. Note, however, that UCW reassignments for natively attached equipment become effective immediately (without a re-IML).

Figure 2 shows the special control-unit and devices requirements to be entered to describe them to the systems for UCW assignments.

The system is generally able to provide better throughput if devices are assigned an unshared UCW and allowed to multiplex; however, this is not always the case. Some shared control units are capable of sustaining only one operation as a time. A shared UCW is quite satisfactory for this type of control unit. Assigning a UCW for each device is wasteful of UCWs. Some control units rely on the characteristics of the selector mode. These control units must have the multiplexing ability of the block-multiplex mode inhibited.

Devices that share a control unit and operate in selector mode can share a common UCW on the channel. The SEL mode bit must be on in the directory entry.

Subchannels and Unit Control Words 13

Type of MPX Channel

Byte Block

- LS -- LS

-- D

- LS - LS - LS - LS - LS - LS - LS

- LD

L s LS - s s s s** s**

L

L

-LS --

Devices

Direct Access Storage Units 2314*-2312,2313,2318,2319 2835*-2305 2841*-2311 3540 3830*-3330,3333,3340,3344,3350 3880*-3330,3340,3344,3350,3370,3375,3380

Magnetic Tape 1/0 2403,2404 2415 2803*-2401,2402,2420 2804*-2401,2402 3411 3411*-3410 3803*-3420

Punched Card 1/0 and Printers 1442 1443 2501 2520 2821*-1403,2540 3203 3811*-3211 3505,3525 3800

Data Security Devices 3848

Displays and Console Printers 2250 2840*-2250 3258*-3251,3255 3272*-3277,3284,3286,3287,3288 3274*-3277,3278,3284,3286,3287,3288,3289 3791*-3277,3284,3286,3287,3288,3793

Magnetic Character and Optical Readers 1255 1287,1288 1419 3881'3886 3890 3895

Communication/Data Acquisition/Process Control 2701 2715*-2791,2793 3704 3705

Paper Tape 1/0 2822*-2671 2826*-1017,1018

Mass Storage System 3850/51

Auxiliary Processor 3838

Audio Response 7770

Key: D

L

s * **

none:

Control Unit/Device capable of operating in data streaming mode. Device requires the block-multiplexer channel to work in selector mode. Device requires a shared UCW. Control unit 3274 Model lA should use an unshared UCW for its single address. Not applicable. Device requires an unshared UCW and the channel working in multiplexer mode.

Figure 2. UCW Assignment Chart


Devices capable of running in block-multiplexer mode can use an unshared UCW for each actual device attached, or one shared UCW for all devices attached to the control unit. Normally, UCWs for the block-multiplexer channel are unshared with SEL mode off.

Some devices, such as the IBM 3272 Control Unit, require one exclusive UCW for each control unit on the channel. Each 3277 attached to that control unit shares that control unit's UCW. The SHARED bit must be on in the directory entry for that UCW.

Magnetic tape devices use a shared UCW and operate in selector mode. Channel-to-channel adapters are treated as control units and require one UCW

for each interface attached to the processor. The use of data streaming mode is independent of whether a UCW is shared or

unshared, or whether operation is in selector or block-multiplexer mode. The use of data streaming mode is still determined by device type.

Subchannels and Unit Control Words I 5

Channel Interference with Processor

Because the processor and the channels share main storage and control storage, execution of processor instructions is slowed by activities on the channels. This is referred to as channel interference with processor or channel and processor interference. For simplicity, these are referred to as interference in this section.

This section defines how to calculate this amount of interference for 1/0 operations. To calculate interference time requires a knowledge of the particular systems 1/0 configurations and of the 1/0 programming involved. To calculate this time involves:

1. Calculation of the total interference time caused by the individual activities on the channel.

2. Calculation of the total interference percentage with the processor by relating the total interference time to a particular time span that is pertinent to the operation; for example, the time taken to execute all or part of an 1/0 operation, or even an entire program. The choice of a particular span of time is up to you.

Calculation of Total Channel/Processor Interference Time The channel activities that can cause channel/processor interference and the duration of those activities are listed in Figures 3 and 4. To calculate total duration of interference:

1. From the figures, list the channel activities and their associated interference times that can occur in the time span chosen.

2. Record the number of times each channel activity (step 1) occurs in this time span.

3. For each channel activity (step 1), multiply the interference time (step 1) by the number of times that the activity occurs (step 2). This product is the duration of interference caused by that activity.

4. Add together the individual duration of interference times (step 3) to obtain the total interference duration.


Channel Mode Channel Activity

Byte Block Selector Multiplex Multiplex Channel

Data Transfer Byte Mode/Byte 16.6 Multi-Byte Mode/Connection 16.6 Burst Mode/64 Bytes 3.0 3.0 3.0

CCW Execution Command Chaining with:

CE and DE Together* 20.8 4.2 4.2 CE and DE Separated* 32.5 15.8 5.8

Data Chaining* 7.0 3.4 3.4 TIC 1. 7 1. 7 1. 7

1/0 Operations Concluding with:

CE and DE Together** 29.8 28.2 24.6 CE and DE Separated** 58.8 61. 5 57.4

PCI Processing 23.0 31. 9 31. 9

IDAW Fetch 1.8 1.8 1.8

* Add time for one IDAW fetch if IDA bit is ON in the new CCW. ** Includes creation time and 1/0 interruption time.

Figure 3. Processor Interference Times in Microseconds (System/370 Mode)

Channel Mode Channel Activity

Byte Block Selector Multiplex Multiplex Channel

Data Transfer Byte Mode/Byte 18.2 Multi-Byte Mode/Connection 18.2 Burst Mode/64 Bytes 3.0 3.0 3.0

CCW Execution Command Chaining with:

CE and DE Together* 23.9 5.8 5.8 CE and DE Separated* 35.6 19.0 7.4

Data Chaining* 10.2 5.0 5.0 TIC 3.3 3.3 3.3

1/0 Operations Concluding with:

CE and DE Together** 29.8 28.2 24.6 CE and DE Separated** 58.8 61. 5 57.4

PCI Processing 23.0 31. 9 31. 9

* Add time for one IDAW fetch if IDA bit is ON in the new CCW. ** Includes creation time and 1/0 interruption time.

Figure 4. Processor Interference Times in Microseconds (ECPS:VSE Mode)

Channel Interference with Processor 17

Figure 5 gives an example calculation of total interference time that is caused by a tape-to-printer operation during which:

1. A 1,000-byte record is read from tape (selector mode on a block-multiplexer channel) by ten data-chained CCWs.

2. The 1,000 bytes are then sent to the printer (byte mode on a byte-multiplexer channel) by ten command-chained CCWs.

3. The example time span is assumed to be the duration of the complete operation.

The duration of interference time does not depend on the data rate of the device but on the characteristics of the channels and on the amount of data being transferred.

Channel Processor Inter- Number of Bytes Duration of Inter fer--Activity Ference Factors or Occurrences ence with Processor

Reading tape in selector mode: Data transfer in burst mode 3.0 usec/byte 1000 bytes/64 48.0 usec

= 16

CCW execution by data chaining 3.4 usec 9 30.6 usec

Concluding of 1/0 operation with CE and DE together 24.6 usec 1 24.6 usec

Writing to printer on byte-multi-plexer channe 1:

Data transfers (in byte mode) 16.6 usec/byte 1000 bytes 16,600 usec

Execution of a CCW by command chaining with CE and DE separated 32.5 usec 9 292.5 usec

Concluding of 1/0 operation with CE and DE separated 58.8 usec 1 58.8 usec

Total duration of interference 17,054.5 usec = 1 7 . 1 ms

Figure 5. Duration of Channel Interference with Processor - (Example Calculation for a Tape-to-Printer Operation in System/370 Mode)

Calculation of Total Channel/Processor Interference Percentage The previous calculations describe how to obtain the actual duration of interference in microseconds over a particular time span. This time figure, divided by the time span used for calculations, yields a percentage of interference.

The significance of the derived percentage of interference depends on the choice of time span. For example, the percentage of interference due to the tape-to-print operation in Figure 5 over a period of one minute is as follows:

[17.1 ms-:- (1 x 60 seconds x 1000 bytes)] x 100 = 0.029%

Conversely, the processor available time is 100%-0.029 = 99.971 %.

The percentage interference figure is the percentage by which the processing of processor instructions is slowed. If the interference is too great for the device being operated (see "Program Overrun"), queues develop for those buffered devices that need service on the byte-multiplexer channel. This results in loss of performance on those devices.


Program Overrun

Assessing Program Overrun

-

Program overrun is a particular effect of channel interference with the processor

(see previous section). Program overrun results from a program being slowed to

such an extent that the program is late in providing real-time service to a device

and, hence, causes incorrect operation of that device. Program overrun must always be considered for those 1/0 operations that

involve high-speed document handling devices, such as the IBM 1419 Magnetic

Character Reader. In program-sort mode, the 1419 reads data into the processor

while the document is passing the read station. Then, before the document

reaches the stacker-select station, the processor must calculate the stacker required

and issue the correct stacker-select command. If the stacker select command

arrives too late (because of program overrun), the document is not stacked

correctly and the channel program stops.

To investigate the possibility of program overrun:

1. Establish the availabie program time, that is, the time during which the program

must perform calculations and issue the command. Call this time A (available

time).

2. Establish the time that the program takes between reading data and issuing the

command. This time can be established by totaling the execution times of the

instructions. (Refer to 11 Instruction Timings 11 in the appropriate Functional

Characteristics manual for your processor Model Group. See the Preface for

publication order numbers.) When establishing this time, take all program

activity into account, including the handling of the 1/0 interruption after the

data is read, and any possible supervisor program activity. Call this time P

(processing time).

3. Establish the maximum possible processor interference time that can be caused

by simultaneous activities on all channels during time A. The calculation of

total channel/processor interference time is described previously in this section.

Call this time I (interference time).

4. Calculate P (processing time) + I (interference time) and compare the result to

A (available time). If P+I is greater than A, program overrun may occur.

Example: Consider the possibility of program overrun with a single-address

1419 and assume that additional channel activities are: (1) a 3330 Disk Storage Device transferring data on one of the channels in burst mode at the rate of 806 kilobytes per second, and (2) a 1442 Card Read Punch using one-byte transfers

and punching EBCDIC characters. For this example, assume that the interference with the processor caused by these

two operations during the available time A is the worst that can occur in this application or that it has the highest interference time I.

Check for program overrun as follows:

1. Establish the available program time A. From IBM 1219 Reader Sorter, IBM 1419 Magnetic Character Reader, GA24-1499, the minimum time available for

giving the stacker-select command is 9.50 milliseconds.

2. Establish the processing time P of the program instruction sequence that

calculates the stacker required and issues the stacker-select command. For the

purpose of this example, assume that the processing time P including possible

supervisor activity, is 8.00 milliseconds.

Channel Interference with Processor 19

3. Establish the total interference time I.

a. Calculate the duration of interference that is caused by the 3330 data transfer within the time span A. At 806 kilobytes per second, the number of bytes transferred in time A (9.50 milliseconds) is:

Number of bytes = (806,000 x 9.50) -:- 1000 = 7,657 bytes

By using the duration of interference on a channel in burst mode to be 3.0 µs/byte (from Figure 3), the total interference in time A caused by transferring 7 ,657 bytes is:

7,657 bytes -:- 64 = 120 64-byte transfers

120 x 3.0 µs per 64 bytes = 360.0 µs of interference time

b. Calculate the duration of interference caused by the 1442 within the time span A. This is assumed to be the execution of one command-chained CCW, and two one-byte data transfers (see Figure 4).

Execution of a CCW with command chaining: 20.8 µs x 1 = 20.8 µs

Data transfer in byte mode: 16.6 µs x 2 = 33.2 µs

Total: 54.0 µs

c. Calculate the total interference time I from steps a and b

360.0 µs (Step a) + 54.0 µs (Step b) = 414.0 µs

or 0.414 ms total interference time

4. Calculate P + I and compare the result with A.

P(8.00 ms) + 1(0.414 ms) = 8.414 ms

or P + I = 8 .414 ms

Because A (9.50 ms) is larger than P + I (8.414 ms), no program overrun can occur.


Concurrent Input/ Output Capabilities

Worst-Case Loads

Overrun

Loss of Performance

Each 1/0 device in operation places a load on its channel facilities. The magnitude of a load depends on a device's channel programming and its data-transfer rate. In this manual, numeric factors are used to relate the loads caused by operation of 1/0 devices to the channel's abilities to sustain concurrent operation of the devices.

One or more numeric factors are specified for each 1/0 device and channel available with the processor. The numeric factors are given in tables in the Appendix and are used in arithmetic procedures for determining whether the operations of specific input/ output configurations are satisfactory.

Several procedures are provided for evaluating a configuration of 1/0 devices for concurrent operation on the channels. For most configurations, using the basic procedures suffices in determining whether operation is satisfactory. Use detailed procedures only for configurations that appear to exceed the processor input/ output capabilities.

The evaluation factors and procedures allow for a worst-case situation, when the most ddemanding devices in the configuration make their heaviest demands on the 1/0 capabilities at the same time. Such a situation may not occur frequently, but this is the situation that the evaluation procedures test. If a particular configuration fails this test, one or more devices can be expected to incur overrun or loss of performance.

Overrun occurs when. a channel does not accept or transfer data within the required time limit. This data loss may occur when the total channel activity initiated by the program exceeds channel capabilities. Depending on the device, the operation may be halted, or data transfer may continue until the end of the block is reached.

An overrun may cause a unit-check indication to be presented to the channel and stored in the, CSW. Chaining, if any, is suppressed, and an I/O interruption condition is generated at the end of the operation. Certain control units, however, can initiate a command-retry sequence without storing a CSW or requiring an 1/0 interruption.

Overrun occurs only on unbuffered 1/0 devices. When buffer service is not provided within required time limits, the device merely waits for channel service. This means a loss of performance.

Conventions for Satisfactory Channel Programs Execution of a channel program causes a load on channel and system 1/0 facilities. Some 1/0 devices require execution of a chain of commands before transfer of a data block. However, the impact of the load caused by a channel program is not a simple function of the number of commands used. The sequence in which particular types of commands appear in a channel program is also a factor.

Control commands are particularly significant to sequencing considerations because they are executed at electronic speed and cause no motion. The command is executed as an immediate operation and provides device end in the initial status byte. When command chaining is specified in such an immediate operation,

Concurrent Input/Output Capabilities 21

channel facilities are not disengaged from the channel program until the chain ends or a command causing mechanical motion or data transfer is executed. Therefore, when immediate operations with device end in the initial status byte are chained together, fetching and execution of the CCWs can cause a heavy load on channel facilities. This load can cause excessive delay in channel service to one or more devices in the 1/0 configuration and result in overrun or Joss of performance. For example, a chain of No-Op commands can contribute heavily to channel loading. Thus, a programming convenience may cause a severe overrun situation for concurrently operating devices.

Data-Chaining Considerations A user can specify data chaining in channel programs, although a channel is able to transfer data at a faster rate, without overrun, when data chaining is not specified. The channel-evaluation procedures and tables in this manual provide guidance in evaluating the effects of data-chaining operations.

Relationship of Conventions and Evaluation Procedures

Scope of Conventions

The evaluation procedures depend on channel programs having command sequences that provide efficient operation of 1/0 devices and that avoid placing unnecessary loads on channel facilities. Channel-programming conventions help 1/0 programmers avoid overrun situations.

Observing channel-programming conventions is fundamental to the selection of an 1/0 configuration that permits satisfactory concurrent operation of 1/0 devices. The channel-programming conventions described are integral to the channel-evaluation procedures. An evaluation indicating satisfactory channel performance is not dependable for channel programs written in violation of the conventions.

1. The conventions relate to the sequence in which certain types of commands can be executed, not to their sequence in main storage.

2. The conventions define four classes of commands and the sequence in which they can be used.

3. The command sequences provided by the conventions differ for different types of devices. Sequences are provided for:

DASO: 2305, 2311, 2314A, 3330, 3333, 3340, 3344, 3350, 3370, 3375, 3380

Tape unit: series 2400, 2420, 3410, 3420 Card units: 1442, 2501, 2520, 2540, 3505, 3525 Printers: 1403, 1442, 3211, 3203 Communication adapters: 2701, 3704, 3705

4. The conventions relate to all the commands in a chain including the CCW addressed by the CAW and the terminating CCW that does not specify any chaining. This is of particular interest to 1/0 programmers working on segments of a single channel program. The conventions still apply when one segment is chained to another segment.

5. The conventions relate only to commands addressed by the CAW that specify chaining.

6. The conventions relate only to the avoidance of overrun; they do not define invalid command sequences that are that are rejected by a channel (such as TIC to TIC), or that are rejected by a control unit. CCW sequences causing command reject are specified in the 1/0 device manuals.


Classes of Commands

The channel-programming conventions in this manual are recommended to 4341

I Processor Model Group 2, 11, and 12 users, particularly in a multiprogramming

environment where a programmer is not aware of the overall load on channel

facilities. Where a programmer controls or has knowledge of all l/0 activity, he

can establish somewhat less restrictive channel programming conventions that are

particularly suited to his application and configuration.

The conventions establish four classes of commands. Commands that always cause

mechanical motion are in one class. The other three classes encompass commands

that are executed at electronic speeds, plus commands that are sometimes executed

at electronic speeds. An example of the latter is rewind, which is executed at

electronic speeds when tape is already at load point. The three classes of

commands having electronic-speed properties differ in the length of time required

for their execution. The conventions for the different devices specify classifications for the specific

commands pertinent to each device. The conventions define the four classifications by the sequence in which they can

precede or follow other commands:

Class A Commands: These commands can be chained in any order, without

restriction. In general, Class A commands always cause mechanical motion.

Class B Commands: Only one Class B command can be chained between two

Class A commands:

permissible command-chaining sequence command-chaining sequence excluded by conventions

A Class B command can be substituted for a Class C or Class D command.

Class C Commands: A Class C command generally appears only once in a channel

program, and then only as the first command in a channel program; therefore, a

Class C command usually appears only in the location specified by CAW:

CAW-+C-+A-+B CAW-+A-+C-+A

permissible program program excluded by conventions

A Class B command can be substituted for a Class C command:

CAW-+B-+A-+B-+A-+B permissible program

Class D Commands: A Class D command can appear only as the last command in

a channel program; it cannot specify any chaining.

CAW-+X-+X-+D CAW-+X-+D-+A

permissible program program excluded by conventions

A Class B command can be substituted for a Class D command.

Some devices have conventions that exclude specific sequences of commands not

excluded by the classifications previously described. Some devices have

conventions that allow a specific command sequence to be substituted for a single

command of a specified class. For detailed information, refer to the publications

that describe the individual devices.


Command Classifications /or I I 0 Devices The following rules define classifications for specific commands used with a particular device. The bit pattern for each command code byte is specified to provide positive identification of commands.

Commands not classified in the following text cannot specify any chaining and can be placed only in the location specified by the CAW. Each such command thus constitutes an entire channel program in which it is the only command. The Sense command is used in this manner for all devices.

Note: For diagnostic or device feature-dependent commands, refer to the device-associated manuals.

Direct-Access Storage Devices: These command classifications are valid for all DASD devices.

Class A commands (any order):

Read xxxx xxlO

Write } Search xxxx xxOl Erase Space Count 0000 1111 Recalibrate 0001 0011 Orient 0010 1011 No-Op 0000 0011

(Class A, 2311 only)

(No-Op can be used only when preceded by a 0001 xxO 1 or 0000 0001 formatting Write.)

Class B commands (not more than one between Class A commands):

TIC

Seek Set Sector Read Sector

xxxx

{ 0000 OOOx 0010

1000 0111 1011 0011

0010 0010

These command chains have the properties of a single Class B command: TIC-Seek xxxx 1000 - { 0000 0111

OOOx 1011

Seek-TIC { 0000 0111} -

OOOx 1011

TIC-Seek-Set Sector

xxxx 1000 - {0000 0111} -OOOx 1011

xxxx 1000

0010 0011

Seek-Set Sector-TIC

0000 0111} OOOx 1011 - 0010 0011 .... xxxx 1000

Note that a Read Sector must be preceded by a Read, Write, or Search command and that a Seek must precede a Set Sector.


Class C commands (first CCW in program). These command chains have the properties of a single Class C command:

Set File Mask 0001 1111 Seek-+Set File Mask-TIC-Set Sector

0000 0111} OOOx 1011 ... 0001 1111 -+ xxxx 1000 0010 0011

Class D commands (last CCW in program): No-Op 0000 0011 (except when preceded

by a formatting Write) Restore 0001 0111 (No-Op on other than

2311 or 2321) Release 0001 0011 (two-channel switch

use only)

Excluded chains:

Search-+ TIC-Write

xOll 0001} xOlO 1001 ... xxxx 1000 ... 0000 xlOl

Data chaining can propagate through a TIC command for gap-only data chaining (see "Data Chaining in Gaps").

Tape Units: Class A commands (any order):

Read xxxx Write Read Backward Forward Space Backspace Write Tape Mark Erase Gap

xxxx xxxx 0011 0010 0001 0001

xxlO xxOl 1100 xlll xlll 1111 0111

Class B commands· (not more than one between Class A commands):

TIC xxxx 1000

Class C commands (first CCW in program): Set Mode xxxx xOll

This command chain has the properties of a single Class C command: Set Mode-+ TIC xxxx xOll-+ xxxx 1000

Class D commands (last CCW in program): Rewind 0000 0111 Rewind and Unload 0000 1111 No-Op 0000 0011 Data Security Erase 1001 0111

Mixed-Mode, Seven-Track Tape Operations: A routine can be used to select a tape unit, set its density mode, and then TIC to a desired channel program:

SIO -+ Set Mode Class C TIC sequence

The conventions require the CCW addressed by the TIC to be Class A.


If the tape applications involve mixed-mode, seven-track operations, the programmer can make provision for placing the proper Set Mode command in the location addressed by the CAW before SIO is issued, or the programmer can begin ..,,,,.,, each channel program addressed by the TIC with an appropriate Set Mode command. This additional Set Mode command violates the convention for Class C commands and causes an additional load on channel facilities.

Card Units:


Read xxxx xx 10 Write xxxx xxOl

Class B commands (not more than one between Class A commands):

TIC xxxx 1000

Class C commands (first CCW in program):

Control xxxx xxll

Class D commands (last CCW in program):

Control xxxx xxll

Printers:


Write xxxx xxOl

Class B command (not more than one between Class A commands):

TIC xxxx 1000


Control xxxx xxl 1


Control 0000 0011

Communication Adapters: Data chaining with or without TIC can be used for these adapters.


~~:e ~ Break Diagnostic write

xxxx xxOl

Read ( Prepare

F:l~stic cead )

xxxx xxlO


Class B commands:

Not applicable


Control* xxxx xxl 1


Control* xxxx xxl 1

• For a communications network of switch-type terminals, these two control commands are Class A:

Disable Enable

0010 1111 0010 0111

Evaluating Heavily Loaded Channels When evaluating the performance of a system susceptible to channel overload conditions, consider the relative ease of restarting an interrupted 1/0 operation. For example, an overrun on a communication line coupling two processors is handled more readily than a read overrun on a card read-punch. Preferential priority can be given to devices that require manual intervention in response to an overrun condition.

Some circumstances may warrant placing devices with heavy load factors on the same block-multiplexer channel, rather than on separate block-multiplexer channels, to reduce the channel load on the system.

Evaluations should not ignore the characteristics of IBM Programming Systems packages. These programs attempt to execute any start 1/0 instruction for which the channel and device are available.


Block-Multiplexer Channel Loading

Overrun Evaluation Consideration When more than one channel is running concurrently, the channels compete for the processor's resources. If any channel is delayed in its bid for the shared resources because the resources are being used by other channels, the channel being delayed may overrun. Block-multiplexer channel operations resulting in overrun cause degradation of the total system because of the time required for error-recovery procedures. Using data chaining with TIC increases the time that a channel needs the shared resources; this can, in turn, reduce the maximum data rate that can be sustained on the channel and can diminish the capabilities of lower-priority channels.

If the program causes data chaining to occur only during gap time, the data chaining does not reduce the maximum data rate that can be sustained on that channel.

Late Command Chaining

Testing for Overrun

In systems that have disk-storage and/or fixed-head storage facilities operating on the lower-priority channels, simultaneous chaining can cause late command chaining on the lower priority channel.

When late command chaining occurs on the facilities with retry capabilities, the channel and control unit cooperate to reissue the late command to repeat the operation after one revolution of the disk. In this case, because it occurs infrequently, system performance is not likely to be significantly affected. Without retry facilities, the channel program ends and an 1/0 interruption condition occurs. This can cause significant performance degradation.

To determine the maximum individual block-multiplexer channel capabilities with various configurations and operating conditions, refer to Appendix C and D. Check to ensure that no device operating under the specified conditions exceeds the maximum data rate specified.


Byte-Multiplexer Channel Loading

The byte-multiplexer channel should not be required to handle a burst-mode 1/0 device that is subject to data overrun. Burst-mode devices (tapes or disk drives) should be attached on a block-multiplexer channel. However, the operation of non-overrunnable buffered 1/0 devices, such as printers, and card readers and punches, is supported in burst mode. Other devices attached to the byte-multiplexer channel are locked out during this period. This can cause overrun when the ability to wait for channel service is exceeded.

Byte-Mode Considerations

Device Load

Device Wait Time

Concurrent operation of 1/0 devices on a byte-multiplexer channel is governed by several variables:

1. Devices vary in data-transfer rates.

2. Devices have buffers varying in capacity.

3. Devices vary in the number and type of CCWs needed for their operation.

4. Combinations of devices on the block-multiplexer channels vary in the interference they cause.

5. The large number of 1/0 devices available for use on a byte-multiplexer channel can be combined in many different ways.

6. Devices in a particular configuration can be physically connected in many different priority sequences.

Worksheet and' factor tables can be used to determine whether a byte-multiplexer channel configuration can run concurrently in a satisfactory manner.

A numeric factor, called a device load, for each byte-mode device represents its load on ,channel facilities. The device load factors are listed in Appendix A for channel 0 and channel 5 (when channel 5 is selected as a second byte-multiplexer channel on the Model Group 12). Appendix B lists the device load factors when channel 4 is selected as a second byte-multiplexer channel.

The values given in Appendix A and B need to be modified if any block-multiplexer channel is active. The Dl factor is used in the following formula to modify the device load:

Device Load (modified) = Device Load + (Dl x n)

where n = number of active block-multiplexer channels.

Appendix A and B lists other factors for use in considering the impact of higher-priority devices on lower-priority devices.

After requesting channel service, a byte-mode device has a fixed time that it can wait for service. If service is provided within this time, the device operates satisfactorily. If, however, the channel does not service the device within the device's wait time, either of two things occurs: if the device is not susceptible to overrun, it continues waiting; if it is, it loses data and causes an 1/0 interrupt condition. For example, if a 1403 Printer on a overloaded byte-multiplexer channel fails to receive data, it waits until service is provided by the channel. The delay does not cause an interrupt condition, nor is a new start I/ 0 instruction required for selecting the 1403. The only effect is a slight reduction in

Byte-Multiplexer Channel Loading 29

performance. If, however, a 1442 Card Read Punch read operation does not receive data service within its wait time, an overrun occurs.

Wait time factors for byte-mode devices are listed in Appendix A and B.

Device Priority on Byte-Multiplexer Channel The priority of devices on a byte-multiplexer channel is established at the time of installation by the sequence in which they are connected to the channel. The cabling facilities provide considerable flexibility in the physical location and logical position of J/0 devices.

Devices may have the priority sequence in which they are attached to the cable (select-out line priority), or the device most remote from the channel may be connected to have highest priority and the device nearest the channel connected to have lowest priority (select-in line priority).

Each device on the byte-multiplexer channel cable may be connected (for selection) either to the select-out line or to the select-in line. Thus, one or the other of the lines is specified in establishing priority for a desired physical layout of devices.

Priority assignments and machine-room layout should be established during the physical planning phase of an installation so that cables for the 1/0 devices can be properly specified.

A major consideration is assigning priority to multiplex-mode devices is their susceptibility to overrun. Devices are identified in this manual as being in one of three classes:

Class 1: Devices subject to overrun; for example, the IBM 2501 Card Reader.

Class 2: Devices that require channel service to be in synchronization with their mechanical operations. For example, the IBM 2540 Card Read Punch has a fixed mechanical cycle. Delay in channel service for such devices ..,,,., usually occasions additional delay because of synchronization lag.

Class 3: Devices that do not require synchronized channel service; for example, the IBM 2660 Display Station with a 2848 Display Control. An IBM 1443 Printer is another device that does not require synchronized channel service; it can begin printing as soon as its buffer is full and line spacing is completed. Any loss of performance by devices in this class is limited to that caused by channel delay in providing service.

Devices in the first class need the highest priority. The devices in the last two classes may operate with reduced performance on an overloaded channel but are not subject to overrun. Their control units have data buffers or an ability to wait for channel service. Devices in the second class, however, should have higher priority than those in the third class.

Within each class, devices are assigned decreasing priority in the order of their increasing wait-time factors; smaller wait-time factors should have higher priority. Wait-time factors are listed in Appendix A and B.

When devices that operate only in burst mode, such as magnetic-tape or disk-storage devices, are attached to the byte-multiplexer channel, they should have lower priority than byte-mode devices. Low-priority devices take longer to respond to selection than do higher-priority devices; a burst-mode device need be selected only once for an operation, but a byte-mode device must be selected for the transfer of each byte, or a short burst, of data.

The control unit determines whether a device operates on the byte-multiplexer channel in burst mode or in byte mode.

Some devices, such as the IBM 2821 Control Unit, can operate on a byte-multiplexer channel in either burst mode or in byte mode, as determined by the setting of a manual switch on the control unit's customer engineer panel. Such


•

devices are assigned priority on the byte-multiplexer channel according to the mode of operation selected.

A byte-multiplexer channel can transfer data most rapidly in burst mode. Where an application uses only class 2 or 3 devices that have the mode choice, improved byte-multiplexer-channel efficiency can be obtained by operating the devices in burst mode.

The support processor and the integrated consoles are connected through the byte-multiplexer channel and have the lowest priority. Other byte-multiplexer channel activity may cause interference with these devices.

Interference from Priority Devices

Priority Loads

Ranges of Wait Times

The byte-multiplexer channel sustains concurrent operations in byte mode by servicing one device at a time.

The operating devices compete for service, and the byte-multiplexer channel services them in the order of their priority.

Devices on the block-multiplexer channels or higher-priority devices on the byte-multiplexer channel can force lower-priority, byte-mode devices to wait for channel service. The former are called priority devices, and the latter waiting devices.

When a priority device forces a waiting device to wait for channel service, the priority device is said to interfere with the lower-priority device.

When there is more than one priority device, each of the priority devices may generate interference. All such interference must be considered in determining whether the waiting device is to receive channel service before its wait time is exceeded.

The test procedures for concurrent operation of byte-mode devices assume that a waiting device has made its request for channel service at the worst possible time; that is, when the priority devices cause maximum interference during the waiting device's wait time.

The channel ordinarily works its way through the interference, and the waiting device is unaffected by the wait. If, however, heavy interference forces the waiting device to wait past its particular wait time, it is subject to overrun.

To evaluate the effect of priority-device interference on a waiting device, a numeric priority load is computed.

Three factors determine a priority load:

1. The control load caused by execution of CCWs including chaining and transfer-in-channel operations.

2. The priority device's data-transfer load.

3. The wait time of the device being evaluated.

Note that because a priority load is a function of wait time, a fixed priority load cannot be established for a priority device; the priority load caused by a priority device must be computed as a function of a particular waiting device's wait time.

The relationship between the interference generated by a priority device (expressed as priority load) and various wait times is shown in Figure 6. The abscissa relates to device wait times. The short wait time shown results in a heavy priority load; the longer wait time falls in a part of the curve showing much less priority load. This curve shows that the impact of a priority device on a waiting device is more intense for a waiting device with a short wait time than it is for a device with a long wait time.


Priority Load

I I I I I

---1------1

I

I I

------1- Short Wait Time

Figure 6. Priority Load Curve

Longer Wait Time

Two factors, called A and B, relate each device's priority load curve to different wait times. The priority load curve was considered in segments related to different time intervals, and an A and a B factor were computed for each curve segment. These factors are used to compute the priority load for a waiting device having a wait time that falls within the range of the interval established for the curve segment.

The values of the A and B factors in Appendix A and B need to be modified if any block-multiplexer channel is active. The Al, A2, and Bl factors are used to modify the given time (T), A, and B factors by the formula:

A (modified) = A + (Al x n) - (A2 x n2)

B (modified) = B + (Bl x n) T (modified) = (difference of modified A values for previous and present

lines) divided by (difference of modified B values for previous and present lines)

n = number of active block-multiplexer channels.

Multiple A and B Factors: The appendixes list the A and B factors for each 4341 Processor Model Group 2, 11, and 12 Class 1 input/ output device.

Some devices have only one set of A and B factors. Other devices have more than one set. Each set has an associated priority time factor that represents the beginning of the time interval over which the A and B factors are effective.

Priority Time Factors: The priority time factors in Appendix A or B are used in the evaluation procedure to identify the A and B factors to be used.

Each waiting device is evaluated on a byte-multiplexer channel worksheet; its wait time is used to select a set of A and B factors for each priority device.

Each set of A and B factors in Appendix A or B has a priority time factor next to it that specifies the beginning of a time interval significant to that set of A and B factors. The range extends from the priority time factor specified for that set to the priority time factor specified for the next set, if any. The end of the last


Priority Load Formula

interval is assumed to be unbounded. For example, a device may have three sets of A and B values that describe the priority load function over three contiguous intervals. Figure 7 shows the priority time factors and A and B factors as they appear in Appendix A for an IBM 1442 Card Read Punch Model Nl reading EBCDIC.

The time priority

intervals are defined time factors:l

by these

0.100 to 0.352

[0.352 to 2.455 2.455 up

A wait time fact 0.915 falls with

Priority Load

Time A B

0.100 9.571 0.000 0.352 4.736 13. 725 2.455 19.177 17.843

or of in this range ...

I+-

,_ ... and this set of A and B factors is significant to a waiting device with a wait time fa~tor of 0.915.

Figure 7. Use of Priority Time Factors

The A and B factors and wait-time factors in Appendix A and B are used in a formula that yields the priority load that occurs when a priority device interferes with a waiting device.

The sum of the B factor and the quotient obtained by dividing the A factor by the wait-time factor of the waiting device is the priority load. The arithmetic looks like this:

(A+wait time) + B = priority load

A-Factor Interference: A-factor interference is caused by channel microprogram activity, such as command chaining, for the priority device. The duration of this type of interference is significant compared with typical wait times. Therefore, the priority load, being expressed as a percentage of wait time, depends on the wait time of the waiting device. For example, if a waiting device's wait time is 0.20 millisecond and the microprogram activity associated with the priority device lasts for 0.10 millisecond, then the priority load is 50 percent. (In the channel evaluation factor tables, the A factors are expressed in milliseconds multiplied by 100. In the foregoing example, the A factor associated with a microprogram activity lasting 0.10 millisecond is therefore: 0.10 x 100 = 10.00.)

Figure 8 shows how A-factor interference varies with the wait time of the waiting device.

B-Factor Interference: B-factor interference is typically caused by data transfers to and from the priority device. As shown by the example in Figure 9, the duration of each data transfer is short compared with typical wait times. . The data transfers occur frequently enough, however, to have a total effect that can be expressed as a percentage interference, namely, the B factor, that is constant for all wait times.


J I Duration* of a block of channel microprogram activity (such as command chaining)

100%------....

80%

60%

Interference

Priority load curve of a priority device whose A factor is 10.00* (with B factor= 0.00)

40%

20%

0 0.10 0.20 0.30

Wait Time (Milliseconds)

*The A factor given in the channel evaluation factor tables (in the appendixes) is the duration of interference in milliseconds multiplied by 100. If the duration of interference is 0.10 millisecond (as shown in the illustration), the A-factor is 0.10 X 100 = 10.00. Thus, for a waiting device having a wait time of 0.2 millisecond, the interference is obtained directly as a percentage:

A-Factor

Wait Time

10.00

0.2 = 50%

0.40 0.50

Figure 8. Example of Priority Device Causing Interference by Command Chaining (A.Factor Interference)

100%

80%

60%

Interference

40%

20%

10%

0

The duration of each data transfer to or from the priority device is small compared with typical wait times. The average percentage interference is the B factor.

/ -.

0.01 0.02 0.03 0.04 Wait Time (Milliseconds)

Priority load curve of a priority device whose B factor is 10.00 (with A factor = 0.00)

Figure 9. Example of Priority Load Curve of a Priority Device Causing Interference by Transferring Data (B~Factor Interference)


-

Priority device

A and B Factors Combined: In actual I/ 0 operations, the pattern of interference tends to be more complex than has been suggested by the example priority load curves in Figures 8 and 9. Usually, the A and B factors are both nonzero, and the total priority load of a priority device is given by:

Priority Load

where the wait time is that of the waiting device.

Multiple A and B Factors: Some devices have only one set of A and B factors but others have more than one set; see the tables in the appendixes. In these tables (as shown in Figure 10), the A and B factors have priority time factors associated with them that show the ranges of wait times (of waiting devices) for which the A and B factors are valid. Figure 10 also shows how to choose the appropriate A and B factors according to the wait time of a waiting device. (The A and B factors for integrated adapters vary, depending upon the channel command. The figures given in the appendixes are representative of worst case interference for a typical channel command sequence.)

These priority time factors . . . . . . . . . . . . . . . . . . . indicate that the

Input/output device Priority load

A B

corresponding A and B

factors are valid for

2520 Card Read Punch Model 81

Read/Punch EBCDIC 0.10 0.00 100.00

3.110 310.98

29.677 4.99

Example 1

When considering the priority load of a 2520 Card Read Punch Model B 1 (that is

reading or punching EBCDIC) upon a waiting device that has a wait time of 1.00 ms,

use the following priority load factors:

A= 0.00 and B = 100.00 (because 1.00 ms is the range 0.10 ms to 3.11 Oms)

Example 2

Similarly, for a waiting device that has a wait time of 10.00 ms, use the following

priority load factors:

A= 310.98 and B = 0.00 (because 10.00 ms is in the range 3.110 ms to 29.667 ms)

Figure 10. Examples Showing How to Choose Priority Load Factors

Previous Load

0.00

10.31

0.10 ms to 3.110 ms

3.110 ms to 29.677 ms

29.677 ms and longer

A waiting byte-mode device may be forced to wait for channel facilities, not only by devices with higher priority, but also by a device with lower priority that is in operation when the waiting device requests channel facilities. This interference is called a previous load and must be added to the priority load caused by priority devices.


Load Sum

Byte-Mode Channel Load Limit

Several load factors relating to byte-mode operations have been described:

Device load (contributed by waiting device) Priority load (contributed by each priority device) Previous load (contributed by a lower-priority device)

These loads added together form a load sum for each waiting device. The load sum represents the total load on system channel facilities under a worst-case condition when:

1. All priority devices are causing maximum priority loads.

2. Any lower-priority device, already in operation, is making maximum demands on channel facilities (previous load).

3. The waiting device places its maximum device load on channel facilities.

A step-by-step procedure for computing load sums is given in "Byte-Mode Evaluation Procedure."

A numeric factor of 100 is considered the byte-mode channel load limit. If a load sum exceeds 100, overrun is indicated during worst-case situations.

Byte-Mode Evaluation Procedure The following step-by-step procedure is used with a IBM 4341 Processor Byte-Multiplexer Channel Preliminary Worksheet (GX24-3746), shown in Figure 11, and a IBM 4341 Processor Channel Load Sum Worksheet (GX24-3670), shown in Figure 12.

Most of the steps call for an entry to be made on a worksheet. Each circled ..,,,, number shown on the worksheets in Figures 13 and 14 refer to the numbered step in the following procedures. For example, a circled number 1 is shown at the top of the worksheet in each of the two spaces that receive the entries called for by step 1. As an additional aid for where entries are made on a worksheet, see Figures 15 and 16 which show worksheets that have been completed for a configuration specified in "Worksheet Example."


~ ~ ~ ~ (") ::r'

i I .,, -.J

(

f --VJ

l ttl

1

I ... (")

i f ~

I

(

~: :f"~ 4341 Byte-Multiplexer Channel Preliminary Worksheet

System Identification:'-------- Block-Multiplexer Channels N values: Nd --- , Np ---

Original Factors

Wait Device Time A B Al Device 01 Time Load

1

2

3

1

2

3

1

2

3

1

2

3

1

2

3

1

2

3

1

2

3

1

2

3

Date:------

A2 Bl Device Previous

Load Load

Modified Factors

Time

1

2

3

1

2

3

1

2

3

1

2

3

1

2

3

1

2

3

1

2

3

1

2

3

(

GX24·3746 U/M 050

File No. 4300-01 Printed in U.S.A.

A B

<.N 31 00

~ .... Cb

txl ...... s:: !V .. Vl <.N

"' .. s ......

'*' "t; ....

n 0 (")

::>"" Cb

§ "' "' 0 ::s .... n ~

t'"' ::>"" 0 "' "' ::s 0.. ::s

~ Vl c: n s ::>""

~ "' .... "' 0 ft ~

"' ::i. ::>"" "' Cb o. ?. (")

"'

'

! : ::f~i: 4341 Processor Channel Load Sum Worksheet GX24-3670 U/M 050 Fol1• No. 4300-01

System Identification

Date

Mp

Priority Devices

Devices at Priority Pos111ons on Bvte-Mult1ple)(er Channel

Waiting Devices (Priority Positions on Byte-Multiplexer Channel)

Device No. Device No Device No .. Device No

Name Name

Wait Priority Load Time Time

Time A B A B A

o I o (ASum) (BSum)

>-----+---+------< A Sum --:---

Device No.

Name

B A

Dev.ce No

Nam>'!

Wait Time

A

Device Nu

Name

Wait Time

~-

Printed in US.A

Device No

f\jam1C

- -+--wo-;-;:---------1 Time

A

I I I I I ~::~~~>me : I : :~:m~I L 8 Sum I I I I I I I I I -+----~---Subtotal•· =

Wall T•me

MP x Prev<OUS f----+---+------l Subtotal Load• f A Sum JI f B Sum)

f----+---+-------l Device Subtotal"• A Sum :

Load - Wa11 Time ----+-------------+----------+----r----, Load Mp x Previous I I

f----+---+------l Sum••• Subtota' Load" (A Sum) fBSum)

Device Subtotal** A Sum: Load - Wait Time

Load Sum•••

MP ' Subtotal

Device

Load

Load Sum•••

Previous

Load"

Subtotal•· _

MP ' Subtotal

Device

Load

Load Sum•••

(A Sum) I (B Sum)

A Sum--:-W;i11 Time

Previous Load"

Subtotal"·

MP ' Subtotal

Dev1CI:' Load

Load Sum•••

Previous Load = 0 for Channel 4 1f the device being evaluated is the lowest priority device.

Subtotal = B Sum + ~ + Previous Load Wait Time

Load Sum = (MP x Subtotal) + Device Load

(

{A Sum) I (8 Sum)

A Sum-;Wait Time

Previous Loao•

Subtotal•• ~

MP ' Subtotal

Device

Load

Load Sum•••

(A Sum) I (B Sum)

A Sum--:---Wa1tT1me =

Previous Load•

Subtotal•• =

MP ' Subtotal

Device

Load

Load Sum•••

(A Sum I (B Sum)

A Sum---;-Wait Time ~

Previous

Load"

Subtotal•• =

MP x Subtotal

Devrce Load

Load Sum••*

(

1 i ~ (")

I ~ i ..., IQ

(

'r1 ~·

= ~ -~ tT1 >< I>' 8 ~ 0 -"tl t ::s ~ ~ 0 .... Pl"

"' =-"' ~ tT1 ::s ~· "'

(

--- -----:::::::::-;-:: 4341 Byte-Multiplexer Channel Preliminary Worksheet

System Identification: © 0 Block-Multiplexer Channels N values: Nd ___ , Np ---

Original Factors

Wait Device Time A B Al Device 01 Time Load

1

3 4 ' -

3

1

2

3

1

2

3

1

2

3

1

2

3

1

2

3

1

2

3

1

4"' ' 3 --

3

Date: Q)

A2 81 Device Previous Load Load

(4) 5/

(4) 5

Modified Factors

Time

1

3

1

2

3

1

2

3

1

2

3

1

2

3

1

2

3

1

2

3

1

-3

' GX24-3746 U/M 050 File No. 4300-01 Printed in U.S.A.

A B

(5)

(5)

"'" 0 ~

- ~ ~ -f"-

"'" ..., "'" I -'ti g " 0 "' -"' 0 .... "t1

(') 5·

l:T e?.

~ ~ [ ~ (') "' [

go ~

I»

!; tI1 ::s

~: s ,.· "'

"'

(

~==~-=~ =~=':'= 4341 Processor Channel load Sum Worksheet GX24-3670 U M 050 Fde No. 4300-01

System Identification G) Waiting Devices (Priority Positions on Byte-Multiplexer Channell

Device No

Date

MP CD .CD

Printed in US.A

Device No

Name Name NamP. Name

Wall Wait Wa11 Wait Priority Load Time

" 'me Priority Devices Time A B A A A A A ""B'

3 A Sum-:-Wait Time"'

@ i-© Previous .(3l1u .. < .. ml Load•

"""OU> ~ I Lood' IS Sumi I ~ I 111 ~·

Load

Devices at Pnority Positions on Bvte-Mult1pJexer Channel

f----+--1--+----l ~~:... 3 :~:~~us (8 Sum)

Subto1a1•• 1'l

(3) -

~ Load Sum•••

Mµ • Subtotal

I~~ Load Sum••• ~

Prev1~us ~ load = · ·""'-

Subtotai••

MP ' Subtotal

I~ Load Sum••• -...........

(A Sum) (8Sum)

~= Previous ~ Lo;:id• = ··~

Sub1ota1•·=

MP ' Subtotat

~ Load Sum•••

Previous Load = 0 for Channel 4 if the device being evaluated is the lowest priority device.

Subtotel = 8 Sum + ~ + Previous Load Weit Time

load Sum • IMP x Subtotell + Device Loed

(

~ fSSum) 1'l (A Sum)

I~-Previous ............... Loa("I• " .. ~

Subtota1••

(A Sum) fBSumJ

~m-C lwa~e=.

Previous Load•

Subtotal••"'

MP x Subtotal

. I~

Load Sum••• ~

l@

3 Subtotat•• = •

MP x Subtotal

I~ Load Sum••• l.3)

(

Preliminary Worksheet Example

This procedure uses the IBM 4341 Processor Byte-Multiplexer Channel Preliminary

Worksheet (GX24-3746).

1. Enter the system identification and the date.

2. This item contains instructions for the calculation of the values Nd, NP and Mp

for the byte-multiplexer channel preliminary and Load Sum worksheets.

In these calculations, Nd is the value used to calculate device load, NP is the

value used to calculate previous load and the A and B factors, and MP is the

value used to calculate previous and prior load on the Load Sum worksheet.

These values are shown on the following charts:

Nd = The number of active block-multiplexer channels, plus 1.0 if the other

byte-multiplexer channel is active, plus .4 for each channel that is data

chaining out of a file gap.

NP = The number of active block-multiplexer channels with high data rate,

plus .4 for each block-multiplexer channel with high data rate that is d~ta

chained when not in a file gap, plus 1.4 if the other byte-multiplexer

channel is also active, plus .6 if the other byte-multiplexer channel is

active and data chaining.

22.5

22.5 - ~ EDR

Where ~ EDR (Effective Data Rate) is the sum of the maximum EDR in

megabytes per sei:ond of each block-multiplexer channel that is not included in

NP. The EDR for each device on a block-multiplexer channel is equal to the

Device Data Rate (DDR) if it is not being data chained, or:

(1 + 256) x DDR CNT

- if the device is being datachained when not in a file gap

Where CNT is equal to the smallest CCW count being used when data

chaining. The starting values for the preliminary and Load Sum worksheets are

obtained from results derived from the following chart. To calculate Nd, NP,

and MP, refer to the following chart (on the next page) and steps a through i:


Sample chart completed for one byte-multiplexer channel on channel 0, and a block-multiplexer channel on channel 4:

( 1 ) ( 2) ( 3) (4) (5) (6) ( 7)

Channel Active Chain Data Byte Mpx Max EDR nd np Used EDR

0 0 0 0 0 0 0 0

1 1 0 0 H 1.0 1.0 0

2 1 1 0 H 1.4 1.4 0

3 1 0 0 H 1.0 1.0 0

4 1 0 0 .333 1.0 0 . 333

5 1 1 0 1.800 1. 4 0 1 .800

5.8 3.4 2.133 22.5

Mp = = 1. 10 (Nd) (Np) ( ~ EDR) 22.5 - 2.133

Blank form for calculations:

( 1 ) ( 2) ( 3) (4) ( 5) (6) (7)

Channel Active Chain Data Byte Mpx Max EDR nd np Used EDR

0

1

2

3

4

5

22.5 Mp = = (Nd) (Np) (~ EDR)

22.5 - ~ EDR

Channel zero is always configured as a byte-multiplexer channel. Additionally, channel 4 (or 5 on the Model Group 12) can be configured as a second byte-multiplexer channel. First enter 'O' in all seven columns of the chart for the byte-multiplexer channel being evaluated.

To arrive at columns 1 through 4 for the remainder of the channels:

a. Column 1 - Enter a 'l' in the 'Active' column for each active channel, and a 'O' in the 'Active' column for each inactive channel. If a zero appears in the 'Active' column, enter 'O' in all remaining columns for that channel.

b. Column 2 - Enter a '1' in the 'Chain Data' column for each active channel that is data chaining when it is not in a file gap; otherwise enter 'O'.


c. Column 3 - Enter a 'l' in the 'Byte-Mpx' column for the other byte-multiplexer channel (channel 0, 4 or 5) if it is being configured as a byte-multiplexer channel; otherwise enter 'O'.

d. Column 4 - Enter a 'O' in the 'Max EDR' column for the other byte-multiplexer channel if it is configured as a byte-multiplexer channel.

The maximum EDR for a channel is determined by calculating an EDR for each device and selecting the maximum value.

Enter an 'H' if the effective data rate (EDR) is high, and will not be used to calculate load; otherwise enter the maximum EDR for each channel.

The EDR for a particular device is determined by the following:

If not data chaining or if data chaining in a file gap.

If data chaining, when not in a file gap.

EDR = Device Data Rate (DDR)

EDR = (1 + 256 ) x DDR CNT

e. Columns 5 and 6 - To calculate a final value for Nd and NP, fill in the nd and np columns for each channel using the following chart:

Co 1 1 Co 1 2 Col 3 Col 4 = Col 5 Col 6

Active Chain Date Byte Mpx Max EDR nd np

1 0 0 H 1.0 1.0 1 1 0 H 1.4 1.4 1 0 0 L 1.0 0.0 1 1 0 L 1.4 0.0 1 0 1 0 1.0 1.4 1 1 1 0 1.4 2.0

f. Column 7 - If the np value in column 6 is 'O', enter the maximum EDR from column 4 in the 'Used EDR' column 7; otherwise enter 'O'.

g. Sum the 'nd', 'np', and 'Used EDR' columns to obtain the Nd, NP, and ~ EDR values.

h. Calculate:

22.5 Mp=

22.5 - ~ EDR

The following table may be- used as a guide to help determine whether to enter 'H' in the maximum EDR column.

Find the column for the byte-multiplexer channel being evaluated. If the accumulative EDR for any number of active block-multiplexer channels exceeds the value given in the table, enter 'H' for those channels; otherwise enter the maximum EDR value.

Note: For accumulative EDR close to the value in the table, it might be necessary to use both methods and then choose the one resulting in the lower Load Sum. MP represents the interference of the block-multiplexer channels on the byte-multiplexer channel resulting from the EDR. However, this interference can never exceed the amount associated with np, which represents the maximum possible interference.

For instance, assume there are a total of three active block-multiplexer channels with respective EDRs of 3.0, 0.4, and 2.5 Mb/second, and the optional byte-multiplexer channel is configured. When channel 0 is evaluated, the accumulative EDR for the first and third block-multiplexer channel exceeds 5.3 Mb/second, 'H' is entered for those channels. However, the accumulative


EDR for all three block-multiplexer channels does not exceed 7.2 Mb/second, so 0. 4 is entered for the second block-multiplexer channel.

ACCUMULATIVE EDR FOR PREVIOUS/PRIORITY LOAD CONSIDERATIONS (a 11 values in Mb/sec)

Active Block Standard Optional Multiplexer Byte MPX Channel Byte MPX Channels

Channels Channel 0 Channel 0 Channel 4

1 3.8 3.0 3.4 2 6.4 5. 3 5.~ 3 8.4 7.2 7. 7 4 10.0 8.6 9.3 5 11. 3 --- ---

If the device being calculated is a 3704 or 3705 operating in the emulation programming mode, go to "IBM 3704/3705 Calculations" to calculate the values for the Load Sum worksheet for those devices.

i. Enter Nd and NP on the preliminary worksheet and MP on the Load Sum worksheet.

3. Arrange the byte-mode devices proposed for simultaneous operation in sequence by priority class (1, 2, and 3).

Class 1 devices that have an inseparable class 2 component should be included with the class 1 devices. Examples of such devices include the 1442 and 2520-Bl.

Enter the class l devices on the worksheet in any order. Although class 2 and 3 devices can be delayed in some worst-case situations,

they are not subject to overrun and need not be include in the overrun evaluation.

4. For each device entered in step 2 that is not a 3705, enter the Wait Time, Device Load, and D l values, and the sets of Time, A, A 1, A2, B, and B 1 values. These values are obtained from Appendix A or B, depending on the multiplexer channel (0, 4 or 5) selected.

If the device is one of the following Class 2 or Class 3 devices, no information is given in Appendix A or B because these devices do not overrun.

1017 1018 1403 1442-N2 1443 2250 2520-B2,B3 2540 2671 2715 3203 3211 3250 3277 3278 3284 3286 3287 3288 3289 3505 3525 3540 3791 3800 3881 3886 3890 3895

The following steps can be performed without referring to the tables. All the needed information is now recorded on the preliminary worksheet.


5. Compute the modified factors for each device by using the following formulas:

Device Load (modified) = Device Load (original) + (Dl x Nd)

Previous Load (modified) = IF there is a slow device on byte-multiplexer channel 0

THEN use [ 10.0 + (1.40 x Np) ] + Wait time

ELSE use [ 7.0 + (1.40 x Np) ] + Wait time

Previous Load (modified) = IF there is a slow device on byte-multiplexer channel 4,

AND it is not the slowest, or last, device,

THEN use [ 9.0 + (1.40 x Np) ] + Wait time

ELSE use [ 6.0 + (1.40 x Np) ] + Wait time

Notes:

11. The slowest, or last, device on channel 4 or 5 (if selected) has a fixed previous value of 0.

2. Slow devices are shown in Table C-2.

AL (modified) = AL (original) + (AIL x NP) - (A2L x Np2)

BL (modified) = BL (original) + (BlL x NP)

Time1 (modified) = Time1 (original)

Time2 (modified) = [A1 (modified) - A2 (modified)] + [B2 (modified) - B1 (modified)]

Time3 (modified) = [A2 (modified) - A3 (modified)] + [B3 (modified) - B2 (modified)l

where:

Nd, NP = effective number of other active channels. WT = wait time of selected device. L = line number 1, 2, or 3 (preliminary worksheet).

6. Proceed to the "Channel Load Sum Worksheet Procedure."

Channel Load Sum Worksheet Procedure This procedure uses the IBM 4341 Processors Channel Load Sum ff'orksheet (GX24-3670).

1. Enter the system identification and date; enter the MP value if not previously entered.

2. Arrange the devices entered on the preliminary worksheet according to increasing wait time. Devices with the smallest wait times are listed first (receive highest priority).

3. Copy the modified factors from the preliminary worksheet to the appropriate areas of this worksheet for each device.

4. For the second and each other waiting device on the byte-multiplexer channel, compare its wait time to the time factors given for each of the byte-multiplexer channel devices with higher priority. Enter the appropriate set of A and B


Worksheet Example

factors (that is, the set that is on the same line with the largest time factor that is less than the wait time).

5. For each byte-multiplexer channel waiting device, add the selected A factors, and enter the A sum and B sum.

6. Divide the A sum by the wait-time factor for the device, and enter the quotient.

7. For each byte-multiplexer channel waiting device, add the B sum, the quotient found in step 6, the previous load, and enter this sum as the subtotal.

8. Calculate MP times the subtotal for each byte-multiplexer channel waiting device, and add this value and the device load and enter this sum as the Load Sum. The load sum must be less than, or equal to, 100 for satisfactory operation of the waiting device.

Channel 1 Channel 2 Channel 3 Channel 4 Byte-Multiplexer

Channel

3330 Disk Storage, no data chaining

2420 Magnetic Tape Unit Model 7 320 Kb/sec, no data chaining 2501 Card Reader Model B2 EBCDIC; 1288 Optical Page Reader, reading formatted alphameric

The completed worksheets (Figures 15 and 16) for the given configuration show satisfactory operation for all byte-mode devices. No load sum exceeds 100.


to '<

'" ~ .a· (';" ><

"' ... (")

~ :::: [ r ~ s·

(JQ

""" -.J

(

"T1 <§" nl Vl

m ><

"' 3 'O (';"

8,

"" a. [f :::: e:; '<

::;:: 0 ... ..,-"' :r "' ~ ~ s: "' N Vl 0 ..... O:i N

"' :::: 0.

"' N 00 00

(

! ;..:;:F;:-1: 4341 Byte-Multiplexer Channel Preliminary Worksheet

System Identification: _______ _ Block-Multiplexer Channels N values: Nd A_ , NP __!}___

Original Factors

Wait Device Device Dl fime , Load

Time A B Al A2

2501-82 I 11.100 1CJ.~ ._O 1.400 0

R£AD!NG .• E'J/5 15.17 3.95 2.~ .4~3 11.46 .863 .026

_EBCDIC 3

12.88 1 .100 15.48 4.65 3.S3G .049 U.NFORMAT7ED 1.00 G.07 1.32. 2 __ ..._

ALPHA MER IC 3

1

2

3

1

f---·--~ >---·-- '--·

2 -+ 3

I

2 I .__ __

3

1

2

3

1

2

3

1

2

3

(

GX24·3746 U/M 050

File No. 4300-01

Date:------Printed in U.S.A.

I Modified Factors

Bl Device Previous Time A B load Load

0 1 .100 16.12. 0

2.305 31.27 17.05 2 .351 8.80 20.68 3

1.241 1 .too 30.41 9.61 13.75 IS.Bo 2

3

1

2

3

1

2

3

1

t 2

3

1

2

3

1

2

3

i 1

2

3

.jl> "r1 00 ~-- @ tJ:I -~ !=7' .jl>

tII ..,, >< .jl> e -"C ~ ..,

0 ("l

0 " "' .... "' !l 0 ..,

::I (") e. :r

~ I» ::I

0 ::I

* g. "' (") :r " :r ~ ~ ~ I»

("l s: (D :i.

I» "' 5· N VI "' 0 ..... ~ N I»

&. I» -N 00 00

(

I::;?~ 4341 Processor Channel Load Sum Worksheet . System Identification

Date

MP

Priority Devices

Devices at Priority Posit1on1 on Byte-Multiplexer Ch1tnnel

Waiting Devices (Priority Positions on Byte-Multiplex• atennal)

4

om .. No2SPJ.8.2. I Dev;ce No .. (~f!J!!. . I [)oy;ce No .•.•...•.• I Dev;ce No •.••....•. 1 Dev;ce No •.•..

Nome Nome Nome N..,.

~: .:~!~ ... ~:, . .l.PO ..... Wait Wllit

P<iorlty Lood Time ............ Time . ............ B A B A B A B A B

0 0 IASuml fBSumJ

8.80120.68 :J0.6 A Sum+ Wait Time = ..

! I.loo j30-1tl9-6t ! 2 I I ! I

..... ou• /'7 ~ R.. J:U) -68 Load• "'.f."!":oV 7A~J (BSuml

Subtotal .... [~.OS :!'~i: =~.8.Q

1----+---+----'~bt:tal .. ('f,,QS ~=~us .. ~'!f.?~ (ASum) (BSumJ

:'.".:~"' 3/. 27 Subtotal .. =1£.08 ~:.~~,;;;. •.

r----+---+----; ~a: ... 48.32. ::,t:tar =f$,.Q8 ~=~us =. IA Sum) IBSumJ

~: /3. 75 Sub101a1••. , . . . . . ::-~i~ = .

~:: ... 58.83 MP x Subtotal •

Device

Lood

Load Sum•••

Previous Load"

Subtoi.1•• ..

MP x Subtotal Device Lood

Load Sum•••

N""e

Wait Tirre

A

(A Sumi I (8 Sumi

A Sum..;. Wait Time=.

Previous ....... Subtota1•• •.

..,, . Subtotal •

Devica

Load

Load Sum•••

Previous Load • 0 for Channel 4 if the device being evaluatld ii the lownt priority d9\IG.

Subtotal • B Sum + ~ + Previous Load Weit Time

Load Sum • (Mp x Subcotall + Dwice L.aed

(

GX24-3870 UIM OIO File No • .aoG-01 Printed in U.S.A.

Device No .•... , •..• I Device No. . ••••.••• I Device No. , •.•. , •..

Nome

Wait Time

A

(A Sumi I (B Sumi

A Sum+ Weit Time• ..

Previous Load.

Subtotei•• •.

Mp x Subtotal •

Device Load

Load Sum•••

Nome

Wait Time

A

IA Sum) I fB Sumi

A Sum..;. Weit Time., .

Previous ....... Subtotei••.

MP x Subtotal

o..,;.,, Load Load Sum•••

Nome

Wait Time

A

IA S'"" I IB Sumi

A Sum..;. Weit Time• .... , .

Previous ....... Subtote1•• •

l·Mp x Subtotal .

OoYiat Load

Load Sum•••

(

IBM 3 704 I 3 705 Calculations This step calculates the values for 3704/3705 devices only. This series of steps provide the information to derive the necessary factors for the Load Sum Worksheet for 3704 and 3705 devices.

The set of numbers in parentheses in the Example Column at the right margin are provided as an example of the values that would be derived if the user has a 3705IJ on byte-multiplexer channel 0, with a type 2 scanner, the device has two lines at 9.6 kilobits per second, and one line at 19.2 kilobits per second.

1. Enter the required data to calculate Wait Time values: Example Column

a. Check the appropriate device. 3704_37051_ 3705Il _(3705II) b. Check the appropriate scanner No. 1 2 3 ( 2 )

(If you have checked scanners 2 and 3, use scanner 2 only. For multiple scanners of the same type, use the most heavily loaded scanner for these calculations.)

c. Check cycle time in microseconds 1.2--1.0 __ .9---( 1.0)

(If you have checked 3704 or 37051, check the value of 1.2).

2. If scanner 3 is checked, enter the buffer length if known, otherwise enter the value 4.

3. If scanner 3 is checked, enter the channel transfer length if known, otherwise enter the value 4.

4. Enter the number of low speed lines (those less than 19,200 bits per second.)

5. Enter the speed of the fastest low speed line in kilobits per second.

6. Calculate the total low speed kilobits per second:

LT = ~ nI I

(where nI is the number of lines at speed I, and I is line speed in kilobits per second, and I is less than 19,200 bits per second.)

7. Enter the number of high speed lines (those equal to or greater than 19,200 bits per second).

8. Enter the speed of the fastest high-speed line in kilobits per second.

9. Calculate the total high speed kilobits per second

HT=~n1I

(where nI is the number of lines at speed I, and I is line speed in kilobits per second, and I is equal to or greater than 19,200 bits per second.)

BFL __ ( 4 )

CTL __ ( 4 )

NL--( 2 )

LM __ ( 9.6)

LT ___ ( 19.2)

NH __ ( 1)

HM __ ( 19.2)

HT __ ( 19.2)

If the above NH = 0, continue to step Al, otherwise go to step A2.


Step At. This step calculates the Wait Time factor when no high speed lines are involved. Refer to Table D-2 for the required values.

From Table D-2, and using the scanner selection, cycle time specification, and buffer load values values from step 1 (previous page), determine the TA, TB, TC and TD values for the following lines:

TA __ _

TB __ _

TC __ _

TD __ _

Calculate TE = Low maximum (LM) +[TC value x Low total (LT)] = TE = __ _

Calculate TF = 10 - [TD value x Low total (LT)] = TF = __ _

Calculate Wait Time = (TA -:- TE) - (TB -:- TF) = Wait Time = __ _

Note: If the device is a 3704, multiply the Wait Time by 0.9.

Proceed to Step B 1.


Step A2. This step calculates the High Wait Time value where high speed lines are involved. Refer to Table D-2 for the required TB and TD values.

From Steps 2 through 9:

Calculate total KB = LT + HT = TKB

To calculate NHM:

If NL = 0, NHM = HM x NH or

If NL ¢ 0, NHM = HM x (NH + 1)

Example Column

TB __ ( .830)

TD ( .0406)

TKB - ___ ( 38.4)

NHM - ___ ( 38.4 )

To calculate the Critical Time (CH) for high speed lines:

If NL = 0, CH = 8 x BFL + NHM or

If NL ¢ 0, CH = 8 x (BFL + 1) + NHM CH - __ ( 1.04 )

To calculate TF:

If NH = 1, TF = 10 or

If NH ¢ 1, TF = 10 - (TD x TKB)

Then calculate:

High Wait Time= CH - (TB+ TF) =

TF =--< 10)

High Wait time =---< .957 ) .

If NL = 0, enter the value for High Wait Time (times 0.9 if 3704 was checked) as the Wait Time at the end of Step A3, and proceed directly to Step Bl.

If NL¢ 0, do step A3 to derive the Low Wait Time value.


Step A3. This step calculates a value of Low Wait Time for a combination of both High Speed and Low Speed lines.

Calculate IA:

IA= .575 x LM-:- (BFL + 1)

Calculate IB:

IB = TKB -:- (16 x BFL)

Calculate IC:

IC = HT -:- (16 x BFL)

Calculate ID:

ID = IB + IC + IA

Calculate IE:

IE = /ID2 - (4 x IB x IC)

Example Column

IA=· ___ ( 1.10)

IB - ___ ( .600)

IC - __ ( .300 )

ID - ___ ( 2.00 )

IE = __ ( 1.81 )

To calculate the Critical Time (CL) for Low Speed Lines:

CL = (ID - IE) -:- (2 x IB x IC) CL - __ ( .528 )

To calculate Low WT:

Low WT = CL - (TB -:- TF) Low WT ---- ( .445 )

For this step final result Wait time, enter the value of High Wait Time calculated in step A2, or this step's calculated Low Wait Time, whichever is the lower value. If the 3704 device had been checked in step Al, multiply

this page final value by 0.9.

Wait Time= ___ ( .445)


,_.. Step Bl. From steps 1 through 6 of the Preliminary Worksheet Example, enter the values for Nd, l'olp and Mp.

Example Column

Nd= ( 4 )

N -p ( 4 )

Mp= ( 1 )

Use the CTL value and the table below to derive the device factor (DP).

CTL=4 CTL=8 I Channel 0 or 5 4.9

Channel 4 4.0 5.5 4.6

Calculate bursts per transfer =

BFL + CTL

To calculate service time (TS).

TS = DP + (1.4 x Np)

To calculate Device Load:

CTL=16 CTL=32 6.7 9.1 5.8 8.2

DP= ( 4.9)

NB-___ ( 1 )

TS =--< 10.5 )

DL = [[DP + (2.4 x Nd)] + [Mp x (NB - 1) x TS]] + WT DL =--< 32.6 )

For channel 0, if there are no slow devices, then:

PL= [7.0 + (1.40 x Np)] +WT

otherwise

PL = [10.0 + (1.40 x Np)] + WT

For channel 4, if there are no slow devices, then:

PL = [6.0 + (1.40 x Np)] + WT

otherwise

PL= [9.0 + (1.40 x Np)] + WT PL== ( 35.1)

Proceed to Step B2.

Byte-Multiplexer Chanttel Loading 53

Step 82. If no lower priority overrunnable devices than the 3704/3705 exist on the channel, priority load factors need not be calculated; proceed directly to the Load Sum Worksheet. Otherwise, calculate the Load Factor values below.

To calculate TSN:

TSN =TS x NB

Example Column

TSN - ___ ( 10.5 )

Calculate and enter the priority Load Factor on the Load Sum Worksheet as follows:

Time2 = .02 x TSN

B2 = (10 x TSN) + [(.l x TSN) +TB]

A2 = Time2 x (100 - B2)

Time1 = 0.100

B1 = 100.000

A1 = 0.000

Timei =---< .210)

B2 = ____ ( 55.85)

A2 = ( 9.27)

Time3 =(NH+ NL) x [(.01 x TSN) + (.1 x TB)] Time3 =---< .564)

B3 = [(LT + HT) x TSN] + (8 x BFL) B3 = ( 12.60)

A3 =---< 33.66 )

Proceed to the 11 Channel Load Sum Worksheet Procedure; 11 enter the Device Load, Previous Load and Priority Load factor values.


> 'O

~ 5. ><" > Ul Ul

(

-l

"' c:r (>'

~ .... (j c:r "' ::; ::; g. tI1 < "' ;:: !':'. 5· ::; '"Ij

"' ("")

0 .... "' l N Ul Ul

"' ::; 0. .... N 00 -.)

(

Nominal Cycle Wait Priority Load Data Rate Time Time Device

Key (kb/sec) (ms) (ms) Load D 1 Time A B Al A2

1255 Magnetic Character Reader

Model 1, 21 lM . 33 120 .680 10. 11 2.824 . 100 17.08 .00 2.800 .000 1. 729 4.57 7.24 .958 .018

34.049 174.93 2.23 2.590 .018

Model 2, 3, 22, 23 lM .50 80 .680 10. 11 2.824 . 100 17.08 .00 2.800 .000 1. 729 4.57 7.24 .958 .018

34.049 136.90 3.35 2.583 .018


Key (kb/sec) (ms) (ms) Load Dl Time A B Al A2

1287 Optical Reader: 1428/ASCSORC Font lM 2.50 Var .40 18. 14 4.80 . 100 25.38 .00 3.920 .000

1 .644 4.60 12.63 .823 .031 1428/ASCSORC Font with lM 2.50 Var . 13 55.83 14.77 . 100 25.38 .00 3. 920 .000

Blank Detection .545 3. 14 40.81 .206 .096 1428 Imprinting lM . 5 Var 2.00 3.63 .96 . 100 24.57 3. 18 3.634 .025 7Bl/ASCSOCR lmprintin lM .40 Var 2.50 2.90 . 77 . 100 24.73 2.56 3.689 .020 Handwritten/Gothic lM .33 Var 1. 50 4.84 1. 28 . 100 24.36 4.01 3.559 .031 Handwritten Verify lM . 20 Var 5.00 1. 45 .38 . 100 25.05 1. 30 3.803 .010

Mode (1287-5) Mark Read 10 Position lM 1. 00 Var 2.00 3.63 .96 . 100 25.46 3.22 3.630 .025 Mark Read 12 Position lM .87 Var 2.30 3. 16 .83 . 100 24.67 2.81 3.667 .022 Alphameric Modes lM 1. 00 Var 1. 00 7.26 1. 92 . 100 23.92 5.77 3.401 .045

(1287-3,4 only} Roll Form lM 2.50 Var .40 18. 14 4.80 . 100 8.61 .00 1. 40 .000

.302 4.60 13.26 .82 .030 Roll Form with lM 2.50 Var .40 18. 14 4.80 . 100 17. 21 .00 2.80 .000

Separate Mark Line .951 4.60 13.26 .82 .030 Command

Roll Form with Blank lM 2.50 Var .20 36.29 9.60 . 100 8.61 .00 1 .40 .000 Detection . 178 3.90 26.53 .53 .060

Roll Form with Blank lM 2.50 Var .20 36.29 9.60 . 100 17. 21 .00 2.80 .000 Detection and Mark .502 3.90 26.53 . 53 .060 Line Command

Key: 1 Device subject to overrun; highest priority required. 2 Device requires synchronized channel service; not subject to overrun; second highest priority required. 3 Device does not require synchronized channel service; not subject to overrun; lowest priority. M Byte mode on byte·multiplexer channel. E Evaluate punching and reading separately by using the wait times, device loads, and previous

loads for the independent operations. Var Variable.

( -> "C

"C B 1 ~ = Q.. -· ~ .000 ?"-

1. 647 .499 ~

< .000

1. 647 .749

~ -= ~ :t. 0 = "rj ~ ~ -0 .... VJ

Bl ...., 0 ....

.000 2.421

n =-.000 8.615

.637

. 513

§ = ~ -.802 Q

.260 0 ....

.645

.562 1. 154

n =-~ .000 = = 2.800 ~ -.000

2.800 (.It -.000 5.600

3: 0

.000 Q.. 5.600 ~ -

~ .... 0 = "C ~

N -

U1 ...., °' I>'

C" .,... - > tc I

~ !'-l .j:O.

(i ..., .j:O.

::>" ..... I>' "t:l = = a ~ g

tr:I "' ~ "' 0 .... i::

(i I>' ::>" 6· I>' = = "I'.1 = " I>' - (') (i 0 ::>"

f I>' .... I>' s .....

N :::!. 00 ~ 00 c;· I>'

= "' p.. ..... .j:O. ..... '°

'



1288 Optical Page Rdr: Formatted Alphameric lM 1.00 Var 1.00 6.07 1. 92 . 100 15.48 4.65 3.536 .049 Unformatted Alphameric lM 1.00 Var 1. 00 6.07 1. 92 . 100 15.48 4.65 3,536 .049 Handwritten/Gothic lM .33 Var 1. 50 4.05 1. 28 . 100 15.74 3.02 3.671 .032 Mark Read 1 lM . 71 Var 1 .40 4.34 1. 37 . 100 15.67 3.44 3.636 .036 Mark read 2 lM .46 Var 2.20 2.76 .87 . 100 15.87 2.21 3.738 .023 Mark read 3 lM .33 Var 3.00 2.02 .64 . 100 15.97 1. 62 3.786 .017 Mark read 4 lM .26 Var 3.80 1. 60 . 51 . 100 16.02 1. 29 3.814 .013 Mark read 5 lM .22 Var 4.60 1. 32 .42 . 100 16.06 1. 06 3.832 .011 Mark read 6 lM . 18 Var 5.40 1. 12 .36 . 100 16.09 . 91 3.845 .009 Mark read 7 lM . 16 Var 6.20 .98 . 31 . 100 16. 10 .79 3.855 .008 Mark read 8 lM . 14 Var 7.00 .87 .27 . 100 16. 12 .70 3.862 .007 Mark read 9 lM . 13 Var 7.80 .78 .25 . 100 16. 13 .63 3.868 .007 Mark read 10 JM . 12 Var 8.60 . 71 .22 . 100 16. 14 .57 3 .873 .006 Mark read 11 lM .11 Var 9.40 .65 .20 . 100 16. 15 .52 3 .877 .005 Mark read 12 lM . 10 Var 10.20 .60 . 19 . 100 16. 15 .48 3.880 .005



1419 Magnetic Character Reader: Single Address Adapter lM 1. 25 32.2 .655 16. 17 4.518 . 100 17.08 .00 2.800 .000

1,600 Documents/ 1.099 4.35 11. 58 .861 .030 Min.

Single Address Adapter lM 1. 25 32.3 .655 16. 17 4.518 . 100 33.99 .00 5.600 .000 With Batch Number- 2.559 4.35 11. 58 .861 .030 ing Feature

Dual Address Adapter lM 1. 25 32.3 .655 16. 17 4.518 . 100 37.29 .00 3.640 .000 without Expanded 2.844 4.35 11. 58 .861 .030 Capability Feature

Dual Address Adapter With Expanded Cap-ability Feature:

- Relative to other lM 1. 25 32.2 .655 16. 17 4.518 . 100 43.64 11. 58 3.651 . 148 Devices in System

- Relative to other lM 1. 25 32.3 .655 16. 17 4.518 . 100 4.35 11. 58 .861 .030 l 4 l 9s in Sys tern

Key: 1 Device subject to overrun; highest priority required. 2 Device requires synchronized channel service; not subject to overrun; second highest priority required. 3 Device does not require synchronized channel service; not subject to overrun; lowest priority. M Byte mode on byte-multiplexer channel. E Evaluate punching and reading separately by using the wait time~, device loads, and previous


(

Bl

1. 211 1. 241

.805

.917

.589

.434

.343

.284

.242

. 211

. 187

. 168

. 152

. 139

. 129

B 1

.000 2.600

.000 2.600

.000 2.600

2.600

2.600

(

f ~ > VI -.J

(

~ ~ :->

g ~ ~ tr:! e. c I» g. =

I "" "" IV I» g_ IV VI 0 -o:l -[ o:l IV

(



1442 Card Reader Model Nl:

Reading EBCDIC lM .53 150 .800 18.60 4.800 .100 9.571 .000 1.400 .000 .352 4.736 13. 725 .813 .031

2.455 19.177 7.843 3.824 .018 Reading Card Image lM 1.07 150 .800 22.68 4.800 . 100 9.571 .000 1.400 .000

.209 5.853 17.800 .721 .031 2.471 24.704 10. 171 3. 772 .018

Punching EBCDIC lM . 12 656 11.000 1. 31 .349 . 100 19.938 .843 3.573 .007 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000

Punching Card Image lM .24 656 11.000 1. 75 .349 .100 24.586 1.226 3.551 .007 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000



2501 Card Reader Model Bl:

Reading EBCDIC lM .80 100 .915 15.47 3.95 .100 10.52 .oo 1.400 .000 .488 4.93 11.46 .863 .026

Reading Card Image lM 1.60 100 .915 18.94 3.95 . 100 10.52 .00 1.400 .000 .291 6.17 14.92 .786 .026

Model B2: Reading EBCDIC lM 1. 33 60 .915 15.47 3.95 . 100 10.52 .00 1.400 .000

.488 4.93 11.46 .863 .026 Reading Card Image lM 2.67 60 .915 18.94 3.95 .100 10.52 .oo 1.400 .000

.291 6. 17 14.92 .786 .026

Key: 1 Device subject to overrun; highest priority required. 2 Device requires synchronized channel service; not subject to overrun; second highest priority required. 3 Device does not require synchronized channel service; not subject to overrun; lowest priority. M Byte mode on byte~multiplexer channel. E Evaluate punching and reading separately by using the wait ti~es, device loads, and previous


(

Bl

.000 2.800 1.600

.000 2.800 1.600

. 179

.000

.000

. 179

.000

.000

Bl

.000 2.305

.000 2.305

.000 2.305

.000 2.305

VI ..., 00 I»

cr' ,... ......

> O:I a:: L. ..

Cl w .. :::.-- I» "ti ::i

::i .... ~ 0

"' trl "' "' < "' a 0 .... "' Cl ~ :::.- c;·

I» ::i ::i

::i 'Tl ~ I»

~ Cl 0 :::.- .... I» "' .... I I»

"' N (;' VI ::i. N ~ 0 r;· "'

'



2520 Card Read/Punch: Reading EBCDIC lM .67 120 1.020 14.74 3.76 . 100 20.87 .00 2.800 .000

1. 540 4.99 10. 31 .889 .023 42.716 277. 33 3.95 56.673 .009 Reading Card Image lM 1. 33 120 1.020 18.05 3.76 . 100 20.87 .00 2.800 .000

1.086 6.28 13.43 .819 .023 42. 733 364.06 5.06 56.648 .009 Punching EBCDIC 2M .67 120 9.000 34.06 5.23 . 100 .00 100.00 .000 .000

3.065 298. 72 2.55 44.637 . 184 Punching Column Binary 2M 1. 33 120 9.000 66. 30 10.20 . 100 .00 100.00 .000 .000 5.967 566.99 4.97 82.707 .703 Reading And Punching 1 ,2M 1. 33 120 .000 .00 .67 . 100 .00 100.00 .000 .000 EBCDIC 3. 110 310.98 .00 47.600 .000

29.677 4.99 10. 31 .889 .023 42.716 174.06 6.35 40.392 .010 Reading and Punching 1 ,2M 2.67 120 .000 .00 1. 33 . 100 .00 100.00 .000 .000 Card Image 6 .011 601 .09 .00 92 .400 .000 42.733 178.31 9.89 27. 179 .017

Key: 1 Device subject to overrun; highest priority required. 2 Device requires synchronized channel service; not subject to overrun; second highest priority required. 3 Device does not require synchronized channel service; not subject to overrun; lowest priority. M Byte mode on byte~multiplexer channel. E Evaluate punching and reading separately by using the wait times, device loads, and previous


(

Bl

.000 2.074

.770

.000 2.074

. 770

.000

.392

.000

.765

.000

.000 2.074 1. 150

.000

.000 1.524

(

f ::s ~ > VI II:>

(

~ (>

~ ~

(")

( i ~ §' 'T1 "" ~ 0

1 -..)

0 -

2701 Data Adapter Unit: IBM Terminal Control I

IBM Terminal Control 11


IBM Telegraph Adapter

Telegraph Adapter Type I

Type 11

World Trade TIY

Synchronous Data Adapter Type I

Type 11 Eight-Bit Code (No Autopolling)

Six-Bit Code (No Autopolling)

Eight-Bit Code (With Autopolling)

Six-Bit Code (With Autopolling)

Bit Data Wait Rate Rate Time (bps) (cps) (ms)

134.5 14.80 63.200 600.0 66.70 14.200

600.0 60.00 14.200

1200.0 120.00 8.300 2400.0 240.00 4.200

75.0 8.33 113.300

45.5 6.00 141. 300 56.9 7.50 113.200 74.2 10.00 86.900

110.0 10.00 85.800

50.0 6.60 128.700 75.0 10.00 85.800

1200.0 150.00 5.800 2000.0 250.00 3.500 2400.0 300.00 2.900

19200.0 2400.00 .360 40800.0 5100.00 . 170

600.0 75.00 25.800 1200.0 150.00 12.900 2000.0 250.00 7.700 2400.0 300.00 6.400 4800.0 600.00 3.200

19200.0 2400.00 .810 40800.0 5100.00 .380 50000.0 6250.00 . 310

230400.0

600.0 100.00 19.200 1200.0 200.00 9.600 2000.0 333.00 5.700 2400.0 400.00 4.800

19200.0 3200.00 .600 40800.0 6800.00 .280 50000.0 8333.00 .230

230400.0

600.0 75.00 14.200 1200.0 150.00 7. 100 2000.0 250.00 4.200 2400.0 300.00 3.500 4800.0 600.00 1.800

600.0 100.00 10.800 1200.0 200.00 5.400 2000.0 333.00 3.200 2400.0 400.00 2.700

( ' Priority Load Device Load D 1 Time A B Al A2 Bl

. 10 .03 .100 9.53 .06 2.517 .000 .017

.42 . 14 . 100 9.50 .27 2.506 .002 .075

.42 . 14 .100 9.51 .24 2.507 .002 .067

.73 .23 .100 9.48 .49 2.495 .003 . 134 1 .43 .46 . 100 9.44 .98 2.470 .007 .269

.05 .02 . 100 9.53 .03 2.518 .000 .009

.04 .01 . 100 9.53 .02 2.519 .000 .007

.05 .02 . 100 9.53 .03 2.518 .000 .008

.07 .02 . 100 9.53 .04 2.518 .000 .011

.07 .02 . 100 9.53 .04 2.518 .000 .011

.05 .01 . 100 9.53 .03 2.519 .000 .007

.07 .02 .100 9.53 .04 2.518 .000 .011

1.04 .33 .100 9.47 .61 2.489 .004 . 168 1. 72 .55 . 100 9.43 1.02 2.468 .007 .280 2.08 .66 . 100 9.41 1. 22 2.457 .008 .336

16.73 5.33 .100 8.60 9. 77 2.018 .068 2.688 35.42 11.29 . 100 7.55 20.76 1 .453 . 144 5.712

.24 .07 .100 9.63 . 31 2.504 .002 .084

.48 . 15 .100 9.60 .61 2.488 .004 .168

.80 .25 .100 9.56 1.02 2.467 .007 .280

.96 .30 .100 9.54 1. 22 2.457 .008 .336 1. 92 .60 . 100 9.42 2.44 2.394 .017 .672 7.59 2.37 . 100 8.71 9. 77 2.014 .068 2.688

16. 18 5.05 . 100 7.65 20.76 1.445 . 144 5.712 19.83 6. 19 .100 7.20 25.44 1.203 . 176 7.000

.32 . 10 .100 9.62 .41 2.499 .003 . 112

.64 .20 . 100 9.58 .81 2.478 .006 .224 1. 08 .34 .100 9.53 1. 36 2.450 .009 .373 1. 28 .40 . 100 9.50 1.63 2.436 .011 .448

10.25 3.20 .100 8.40 13.02 1.846 .090 3.584 21.96 6.86 . 100 6.99 27.67 1.087 . 192 7.616 26.73 8.35 . 100 6.38 33.91 .764 .235 9.333

.43 . 14 .100 9.63 . 31 2.504 .002 .084

.87 .27 . 100 9.60 .61 2.488 .004 . 168 1.46 .46 .100 9.56 1.02 2.467 .007 .280 1. 76 .55 .100 9.54 1. 22 2.457 .008 .336 3.42 1.07 . 100 9.42 2.44 2.394 .017 .672

.57 . 18 . 100 9.62 .41 2.499 .003 . 112 1. 14 .36 .100 9.58 .81 2.478 .006 .224 1. 92 .60 . 100 9.53 1. 36 2.450 .009 .373 2.28 . 71 .100 9.50 1.63 2.436 .011 .448

O'I >-1 0 ., er ;:;-

63 > ~ °' """ <1 w

""" ::>" ..... ., "ti ::s ::s 8 (E.. (')

trl "' "' < "' ~ 0 .... :::: <1 ~ ::>" s· .,

::s ::s 'T"l ::I

(E.. ., <1 8 ::>" ....

"' .,

I .... ., (') _,

'" _,

::i. _, "' 0 g. "'

(

Data Wait Priority Load Rate Time Device (cps) (ms) Load D 1 Time A B Al A2

7770 Audio Response Unit: 8 Lines 12 9.02 .70 . 213 . 100 6.559 4.45 1. 275 .013 16 Lines 12 4.49 1.40 .427 . 100 6.559 4.45 1. 275 .013 24 Lines 12 2.99 2. 10 .643 . 100 6.559 4.45 1. 275 .013 32 Lines 12 1.48 4.25 1. 300 . 100 6.559 4.45 1. 275 .013 40 Lines 12 1. 48 4.25 1. 300 . 100 6.559 4.45 1. 275 .013 48 Lines 12 1. 48 4.25 1.300 . 100 6.559 4.45 1. 275 .013

Key: 1 Device subject to overrun; highest priority required. 2 Device requires synchronized channel service; not subject to overrun; second highest priority required. 3 Device does not require synchronized channel service; not subject to overrun; lowest priority . M Byte mode on byte-multiplexer channel . E Evuluate punching and reading separately by using the wait times, device loads, and previous loads for the independent operations. Var Variable.

(

Bl

.909

.909

.909

.909

.909

.909

(

i e: >< c:l

°' -

(

~ (i"

c:l I -n

[ ~ ~-0

~ 0

I N UI UI

~ -N 00 .....

(



1255 Magnetic Character Reader

Model 1, 21 lM .33 120 .680 9.27 2.824 . 100 15.22 .00 2.800 .000 1. 741 4.07 6.40 .977 .018

34.044 154.65 1.98 2.609 .018

Model 2, 3, 22, 23 lM .50 80 .680 9.27 2.824 .100 15.22 .00 2.800 .000 1. 741 4.07 6.40 .977 .018

34.044 121.01 2.96 2.602 .018



1287 Optical Reader: 1428/ASCSORC Font lM 2.50 Var .40 16.64 4.80 . 100 22.80 .00 3.920 .000

1.655 4. 15 11. 27 .856 .031 1428/ASCSORC Font with lM 2.50 Var . 13 51. 21 14. 77 .100 22.80 .00 3.920 .000

Blank Detection .547 3.00 36.20 .309 .096 1428 Imprinting JM .5 Var 2.00 3. 33 .96 .100 22. 15 2.83 3.664 .025 7Bl/ASCSOCR lmprintin JM .40 Var 2.50 2.66 .77 . 100 22.28 2.28 3.714 .020 Handwritten/Gothic JM .33 Var 1.50 4.44 1. 28 .100 21. 99 3.57 3.597 .031 Handwritten Verify JM .20 ~ar 5.00 1.33 .38 .100 22.54 1. 15 3.815 .010

Mode (1287-5) Mark Read 10 Position lM 1.00 Var 2.00 3.33 .96 . 100 22. 15 2.87 3.660 .025 Mark Read 12 Position lM .87 Var 2.30 2.89 .83 .100 22.23 2.50 3.694 .022 Alphameric Modes lM 1.00 Var 1.00 6.66 1. 92 .100 21.63 5. 13 3.455 .045

(1287-3,4 only) Roll Form JM 2.50 Var .40 16.64 4.80 .100 7.62 .00 1.400 .000

.295 4. 15 11. 76 .860 .030 Rol 1 Form with JM 2.50 Var .40 16.64 4.80 . 100 15.23 .00 2.800 .000

Separate Mark Line .942 4. 15 11. 76 .860 .030 Command

Roll Form with Blank lM 2.50 Var .20 33.29 9.60 . 100 7.62 .00 1.400 .000 Detection . 171 3.60 23.53 .590 .060

Roll Form with Blank lM 2.50 Var .20 33.29 9.60 .100 15.23 .00 2.800 .000 Detection and Mark .495 3.60 23.53 .590 .060 Line Command

Key: 1 Device subject to overrun; highest priority required. 2 Device requires synchronized channel service; not subject to overrun; second highest priority required. 3 Device does not require synchronized channel service; not subject to overrun; lowest priority. M Byte mode on byte-multiplexer channel. E Evaluate punching and reading separately by using the wait times, device l_oads, and previous


(

Bl

~ 1 a=

.000 = . 1.647

.499

.000 1.647

.749

~ e.. = ! g "'!l!j

a !

Bl ~ = .000

2.429 .000

8.615 l 3:

.638

.513

.803

.260

e. ~ '2. tl

.645

.563 ~ 1.156 n

=" .000

2.800 .000

2.800

i !1. ~

.000 5.600

.000 5.600

>-l °' ~ N er ,..

ti1 ~ ~ N """ (") ..,,

;:r """ ~ - ::s "Cl ::s .... !!. 0

()

tr! ('!)

< "' e.:. "' 0 c ....

(") ~ s· ;:r ::s ~ ::s 'T1 ::s ~ !!. s (") .... ;:r

1 ~ .... P.

N ('!)

00 00

!:!. C.

~ ::s ()

"' c.. -""" ..... \0

(



1288 Optical Page Rdr: Formatted Alphameric lM 1. 00 Var 1. 00 5.35 1. 92 . 100 12.67 3.83 3.606 .049 Unformatted Alphameric lM 1. 00 Var 1. 00 5.35 1. 92 . 100 12.67 3.83 3.606 .049 Handwritten/Gothic lM .33 Var 1. 50 3.57 1. 28 . 100 12.85 2.40 3.716 .032 Mark Read 1 lM . 71 Var 1. 40 3.82 1. 37 . 100 12.80 2.83 3.688 .036 Mark read 2 JM .46 Var 2.20 2.43 .87 . 100 12.93 1. 82 3. 771 .023 Mark read 3 lM . 33 Var 3.00 1. 78 .64 . 100 13.00 1. 34 3.810 .017 Mark read 4 JM .26 Var 3.80 1 .41 . 51 . 100 13.03 1.06 3.833 .013 Mark read 5 JM .22 Var 4.60 1. 16 .42 . 100 13.06 .88 3.848 .011 Mark read 6 lM . 18 Var 5.40 .99 .36 . 100 13.07 .75 3.859 .009 Mark read 7 lM . 16 Var 6.20 .86 . 31 . 100 13.09 .65 3.867 .008 Mark read 8 JM . 14 Var 7.00 .76 .27 . 100 13. 10 .58 3 .873 .007 Mark read 9 lM . 13 Var 7.80 .69 .25 . 100 13. 10 .52 3.878 .007 Mark read 10 lM . 12 Var 8.60 .62 .22 . 100 13. 11 .47 3.881 .006 Mark read 11 lM . 11 Var 9.40 .57 .20 . 100 13. 12 .43 3.885 .005 Mark read 12 JM . 10 Var 10.20 .52 . 19 . 100 13. 12 .40 3.888 .005


Key (kb/sec) (ms) (ms) Load D 1 Time A B Al A2 1419 Magnetic Character Reader: Single Address Adapter JM 1. 25 32.2 .655 14.83 4.518 . 100 15.22 .00 2.800 .000 1,600 Documents/ 1. 105 3. 91 10.24 .891 .030 Min. Single Address Adapter JM 1. 25 32.3 .655 14.83 4.518 . 100 30.27 .00 5.600 .000 With Batch Number- 2.575 3. 91 10.24 .891 .030 ing Feature Dual Address Adapter lM 1. 25 32.3 .655 14.83 4.518 . 100 34.83 .00 3.640 .000 without Expanded 3.020 3.91 10.24 .891 .030 Capability Feature Dual Address Adapter

With Expanded Cap-ability Feature:

- Relative to other JM 1. 25 32.2 .655 14.83 4.518 . 100 40.96 10.24 3.824 . 148 Devices in System - Relative to other JM 1. 25 32.3 .655 14.83 4.518 . 100 3.91 10.24 .891 .030 1419s in System

Key: 1 Device subject to overrun; highest priority required. 2 Device requires synchronized channel service; not subject to overrun; second highest priority required. 3 Device does not require synchronized channel service; not subject to overrun; lowest priority. M Byte mode on byte-multiplexer channel. E Evaluate punching and reading separately by using the wait times, device loads, and previous loads for the independent operations. Var Variable.

(

B 1

1. 244 1. 244

.806

.918

.589

.434

.343

.284

.242

. 211

. 187

. 168

. 152

. 140

. 129

B 1

.000 2.600

.000 2.600

.000 2.600

2.600

2.600

(

~ 'O

"' ~ Ill

0-, w

(

;3 r::r 10 i;c w

("') ::r "' Q Q

~

tTl .,, "' 2" ~ 5· Q ..,, "' (")

0

l .j:>. .j:>. N

"' Q 0.. N v. 0

Ill

"' Q 0..

Ill N

(

Nomi na 1 Cycle Wait Priority Load Data Rate Time Time Device


1442 Card Reader Model Nl:

Reading EBCDIC lM .53 150 .800 17.03 4.800 . 100 8.521 .000 1.400 .000 .350 4.270 12. 150 .848 .031

2.449 17.020 6.943 3.844 .018 Reading Card Image lM 1. 07 150 .800 21. 10 4.800 . 100 8.521 .000 1. 400 .000

. 190 5.437 16.225 .757 .031 2.465 22. 577 9.271 3.792 .018

Punching EBCDIC lM . 12 656 11 .000 1. 21 .349 . 100 17.782 .752 3.581 .007 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000

Punching Card Image JM .24 656 11.000 1 .64 .349 . 100 22.443 1. 134 3.558 .007 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000



2501 Card Reader Mode 1 B 1:

Reading EBCDIC lM .80 100 . 915 14.30 3.95 . 100 9.56 .00 1. 400 .000 .494 4.48 10.28 .890 .026

Reading Card Image lM 1. 60 100 . 915 17. 77 3.95 . 100 9.56 .00 1 .400 .000 .276 5.76 13.75 .812 .026

Model B2: Reading EBCDIC lM 1. 33 60 .915 14.30 3.95 . 100 9.56 .00 1. 400 .000

.494 4.48 10.28 .890 .026 Reading Card Image lM 2.67 60 . 915 17. 77 3.95 . 100 9.56 .00 1 .400 .000

.276 5.76 13.75 . 812 .026

Key: 1 Device subject to overrun; highest priority required. 2 Device requires synchronized channel service; not subject to overrun; second highest priority required. 3 Device does not require synchronized channel service; not subject to overrun; lowest priority. M Byte mode on byte-multiplexer channel. E Evaluate punching and reading separately by using the wait times, device loads, and previous


(

B 1

.000 2.800 1.600

.000 2.800 1 .600

. 179

.000

.000

. 179

.000

.000

B 1

.000 2.305

.000 2.305

.000 2.305

.000 2.305

0\ ~ -!>- "" ~ - tc tc

.):.. :::: -!>-

('j (;>

-!>- ::r ...... "" "ti i:l i:l .... ~ 0

0 tn " "' < "' E-0 ....

('j "" ::r 6· "" i:l i:l

'Tl i:l ~ "" 0 ('j 0 ::r .., "" 1 .... "" 0 fD VI ;::!. N

"' 0 g. "'

(



2520 Card Read/Punch: Reading EBCDIC JM .67 120 1.020 13.63 3.76 . 100 18.95 .00 2.800 .000

1. 557 4.54 9.26 . 913 .023 42.710 248.88 3.53 56.680 .009 Reading Card Image lM 1. 33 120 1 .020 16.93 3.76 . 100 18.95 .00 2.800 .000

1.058 5.86 12.38 .843 .023 42. 727 335.62 4.66 56.654 .009 Punching EBCDIC 2M .67 120 9.000 31. 35 5.23 . 100 .00 100. 00 .000 .000

2.822 275.55 2.35 44.828 . 184 Punching Column Binary 2M 1.33 120 9.000 61 .06 10.20 . 100 .00 100.00 .000 .000 5.495 524.33 4.58 83.429 .703 Reading And Punching 1,2M 1. 33 120 .000 .00 .67 . 100 .00 100.00 .000 .000 EBCDIC 2.863 286.26 .00 47.600 .000

30.440 4.54 9.26 . 913 .023 42.710 153.74 5.76 40.405 .010 Reading and Punching 1,2M 2.67 120 .000 .00 1. 33 . 100 .00 100.00 .000 .000 Card Image 5.536 553.57 .00 92.400 .000 42. 727 164.22 9. 11 27. 196 .017

Key: 1 Device subject to overrun; highest priority required. 2 De~ice requires synchronized channel service; not subject to overrun; second highest priority required. 3 Device does not require synchronized channel service; not subject to overrun; lowest priority. M Byte mode on byte-multiplexer channel. E Evaluate punching and reading separately by using the wait times, device loads, and previous


( ..

B 1

.000 2.074

.770

.000 2.074

. 770

.000

.392

.000

.765

.000

.000 2.074 1. 150

.000

.000 1. 524

(

> "O

~ [ ~-

~

0-. VI

(

~ ~ ~ ~

('}

( tr! <

I ::s

~ 0

1 .... 0 .....

2701 Data Adapter Unit: IBM Terminal Control I



IBM Telegraph Adapter

Telegraph Adapter Type I

Type 11

World Trade TTY

Synchronous Data Adapter Type I

Type 11 Eight-Bit Code (No Autopolling)

Six-Bit Code (No Autopolling)

Eight-Bit Code (With Autopolling)

Six-Bit Code (With Autopolling)

Bit Data Wait Rate Rate Time (bps) (cps) (ms)

134.5 14.80 63.200 600.0 66.70 14.200

600.0 60.00 14.200

1200.0 120.00 8.300 2400.0 240.00 4.200

75.0 8.33 113.300

45.5 6.00 141. 300 56.9 7.50 113.200 74.2 10.00 86.900

110.0 10.00 85.800

50.0 6.60 128.700 75.0 10.00 85.800

1200.0 150.00 5.800 2000.0 250.00 3.500 2400.0 300.00 2.900

19200.0 2400.00 .360 40800.0 5100.00 . 170

600.0 75.00 25.800 1200.0 150.00 12.900 2000.0 250.00 7.700 2400.0 300.00 6.400 4800.0 600.00 3.200

19200.0 2400.00 .810 40800.0 5100.00 .380 50000.0 6250.00 .310

230400.0

600.0 100.00 19.200 1200.0 200.00 9.600 2000.0 333.00 5.700 2400.0 400.00 4.800

19200.0 3200.00 .600 40800.0 6800.00 .280 50000.0 8333.00 .230

230400.0

600.0 75.00 14.200 1200.0 150.00 7. 100 2000.0 250.00 4.200 2400.0 300.00 3.500 4800.0 600.00 1. 800

600.0 100.00 10.800 1200.0 200.00 5.400 2000.0 333. 00 3.200 2400.0 400.00 2.700

( (

Priority Load Device Load Dl Time A B Al A2 Bl

.08 .03 . 100 7 .61 .05 2.518 .000 .017

.37 . 14 . 100 7.59 .22 2.509 .002 .075

.37 . 14 . 100 7.60 .20 2.510 .002 .067

.63 .23 . 100 7.58 .39 2.500 .003 . 134 1. 25 .46 . 100 7.55 .79 2.480 .007 .269

.05 .02 . 100 7.61 .03 2.519 .000 .009

.04 .01 . 100 7.61 .02 2.519 .000 .007

.05 .02 . 100 7.61 .02 2.519 .000 .008

.06 .02 . 100 7.61 .03 2.518 .000 .011

.06 .02 . 100 7.61 .03 2.518 .000 .011

.04 . 01 . 100 7.61 .02 2.519 .000 .007

.06 .02 . 100 7.61 .03 2.518 .000 . 011

.90 .33 . 100 7.57 .49 2.495 .004 . 168 1. 50 .55 . 100 7.55 .82 2.478 .007 .280 1.81 .66 . 100 7.54 .99 2.470 .008 .336

14.56 5.33 . 100 7.01 7.90 2. 116 .068 2.688 30.83 11 .29 . 100 6.33 16.78 1.662 . 144 5.712

.21 .07 . 100 7.72 .25 2.507 .002 .084

.42 . 15 . 100 7.70 .49 2.495 .004 . 168

.70 .25 . 100 7.67 .82 2.478 .007 .280

.84 .30 .100 7.66 .99 2.469 .008 .336 1.68 .60 . 100 7.59 1. 97 2.418 .017 .672 6.63 2.37 . 100 7. 13 7.90 2. 113 .068 2.688

14. 13 5.05 . 100 6.44 16.78 1 .655 . 144 5.712 17.32 6. 19 . 100 6. 15 20.56 1.460 . 176 7.000

.28 . 10 . 100 7. 71 .33 2.503 .003 . 112

.56 .20 . 100 7.69 .66 2.486 .006 .224

.94 .34 . 100 7.65 1. 10 2.464 .009 . 373 1. 12 .40 . 100 7.64 1. 32 2.452 . 011 .448 8.95 3.20 . 100 6.92 10.53 1. 997 .090 3.584

19. 17 6.86 . 100 6.01 22.37 1. 367 . 192 7.616 23.34 8.35 . 100 5.62 27.41 1. 107 .235 9.333

.38 . 14 . 100 7. 72 .25 2.507 .002 .084

.76 .27 . 100 7.70 .49 2.495 .004 . 168 1. 28 .46 . 100 7.67 .82 2.478 .007 .280 1. 53 .55 . 100 7.66 .99 2.469 .008 .336 2.98 1. 07 . 100 7.59 1. 97 2.418 .017 .672

.50 . 18 . 100 7.71 .33 2.503 .003 . 112

.99 .36 . 100 7.69 .66 2.486 .006 .224 1. 68 .60 . 100 7.65 1. 10 2.464 .009 .373 1. 99 . 7 ~ . 100 7.64 1. 32 2.452 . 011 .448

a-. ..., a-. "' ~ 63 t:C

I ~ ?'

""" (') w

""" :::>" ..... "' ::i "ti ::i ..., g. 0

"' tT1 "' "' < "' e:. 0 ...,

~-(') :::>"

"' ::i ::i "r1 ::i

"' "' - "' (') 0 :::>" ..., "' T ..., "' ~ -.J

-.J -.J ::J. 0 "' o.

"' "'

(

Data Wait Priority Load Rate Time Device (cps) (ms) Load D 1 Time A B Al A2

7770 Audio Response Unit: 8 Lines 12 9.02 .62 .213 . 100 5.651 3.81 1. 293 .013

16 Lines 12 4.49 1. 24 .427 . 100 5.651 3.81 1. 293 .013 24 Lines 12 2.99 1. 87 .643 . 100 5.651 3.81 1. 293 .013 32 Lines 12 1.48 3.79 1. 300 . 100 5.651 3. 81 1. 293 .013 40 Lines 12 1. 48 3.79 1. 300 . 100 5.651 3.81 1. 293 .013 48 Lines 12 1.48 3.79 1. 300 . 100 5.651 3.81 1. 293 .013

Key: 1 Device subject to overrun; highest priority required. 2 Device requires synchronized channel service; not subject to overrun; second highest priority required. 3 Device does not require synchronized channel service; not subject to overrun; lowest priority. M Byte mode on byte-multiplexer channel. E Evaluate punching and reading separately by using the wait times, device loads, and previous


(

B 1

.909

.909

.909

.909

.909

.909

(

Appendix C. Device Data Rates

Device and Model

2250-3 2305-2 2311-1 2314-A or B 3330 3340 3344 3350 3370 3375 3380 3848

2401, 2402, 2403, 2404-1

2401, 2402, 2403, 2404-2

2401, 2402, 2403, 2404-3

2401, 2402, 2403, 2404-4

2401, 2402, 2403, 2404-5

2401, 2402, 2403, 2404-6

Device and Model

2415

2420-5 2420-7 3410-1

3410-2

3410-3

Density

200 556 800

1600 1600 1600 200 556 800

1600 200 556 800

1600 200 556 800

1600

Table C-1. Data Rates by Device

Nominal Speed Command Retry on Command Retry on Density (Megabytes/Second) Command Overrun? Data Overrun?

0.526 1.500 o. 156 0.312 0.806 0.885 0.885 1. 198 1. 859 1.859 3.000 3.000

200 556 800 200 556 800 200 556 800 800

0.0075 or 0.0056* 0.0208 or 0.0156* 0.0300 or 0.0225* 0.0150 or 0.0113* 0.0417 or 0.0313* 0.0600 or 0.0450* 0.0225 or 0.0169* 0.0625 or 0.0469* 0.0900 or 0.0675*

1600 800

1600 800

1600

0.0300 0.0600 0.0600 0. 1200 0.0900 0.1800

Nominal Speed Device (Megabytes/Second) and Model

0.0038 3420-3 0.0104 0.0150 0.0300 3420-4 0. 1600 0.3200 3420-5 0.0025 0.0070 0.0100 3420-6 0.0200 0.0050 3420-7 0.0139 0.0200 0.0400 3420-8 0 .0100 0.0278 0.0400 0.0800

No Yes No No Yes No No Yes Yes Yes Yes No **

No Yes No No Yes No No Yes Yes Yes Yes No **

•Data conversion in effect.

**3848 is fully buffered, not subject to overrun.

Density

556 800

1600 1600 6250

556 800

1600 1600 6250

556 800

1600 1600 6250

Nominal Speed (Megabytes/Second)

0.0417 0.0600 0. 1200 0. 1200 0.4688 0.0695 0. 1000 0.2000 0.2000 0.7803 0. 1112 0. 1600 0.3200 0.3200 1.2500

Appendix C 67

CONTROL UN IT ( S) DEVICE TAG INSTANTANEOUS DATA RATE SPEED (Mb/sec)

Direct Access Storage Devices: 3540 slow Note 1

Punched Card 1/0 and Printers: 1442 slow Note 1 1443 slow Note 1 2501 slow Note 1 2520 slow Note 1

2821 1403 slow 0.200 2821 2540 slow 0. 166

3203 fast 0.598 3811 3211 slow 0.288

3505 fast EBCDIC: 0. 167 I card image 0.350 3525 fast EBCDIC: 0. 167 I card image 0.350 3800 fast 0.235

Display and Console Printers: 3272 fast 0.650 3274-lA fast Note 1 3274-lB, ID fast 0.650 3791 3793 fast Note 1

Magnetic Character and Optical Reader: 1255 slow Note 1

1287 slow Note 1 1288 fast Note 1 1419 slow Note 1 3881 slow 0. 115 3886 fast 0. 126 3890 slow 0.200 3895 fast Note 2

Communication/Data Acquisition/ Process Control: 2701 fast Note 1

2715 2791 slow Note 2 2715 2793 slow Note 2

3704 fast Note 2 3705 fast Note 1

Paper Tape 1/0: 2822 2671 slow Note 1 2826 1017 slow Note 2 2826 1018 slow Note 2

3850/51 fast Channel Speed (3 Mb/sec or 2 Mb/sec)

Audio Response Unit: 7770 slow Note 2

CHAN-to CHAN fast Channel Speed (3 Mb/sec or 2 Mb/sec)

ADAPTER

Notes:

1. Negligible for EDR consideration.

2. Not applicable for EDR consideration.

Table C-2. Data Rates by Device


Appendix D. Miscellaneous Tables

SECTION 1 Channel 1 Channel 2 Channel 3 Channel 4 Channe 1 5

No Data Chaining I 3.0 Mb 3.0 Mb 2.0 Mb 2.0 Mb * 2.0 Mb •

SECTION 2 Data Chaining

Sma 1 lest CCW Count

1 6Kb 6Kb 5Kb 4Kb 4Kb 2 13 11 10 9 8 4 27 23 20 18 16 8 54 46 40 36 32

16 104 89 77 69 62 32 193 166 143 129 115 64 340 291 251 226 201

128 546 467 402 363 322 256 784 671 576 520 461 512 1002 857 734 664 588

1024 1164 995 850 771 681 2048 1266 1082 923 838 740 4096 1325 1132 965 876 773 8192 1356 1158 987 897 791

16384 1372 1172 999 907 800 32768 1380 1179 1005 913 805 65535 1384 1182 1008 916 807

• 3.0 Mb on Model Group 12

Table D-1. Data Chaining Rates for Block-Multiplexer Channels

Appendix D 69

BFL TA TB TC TD

Scanner 1 4 8.70 .996 . 124 .35000

Scanner 2 CT = 1. 2 4 8. 70 .996 . 136 .04880 CT = 1.0 4 8. 70 .830 . 136 .04060 CT = 0.9 4 8. 70 .747 . 136 .03660 ..

Scanner 3 CT = 1. 2

4 8. 70 1. 420 . 136 .01550 8 15.70 1. 870 . 122 .00939

16 29.60 2.780 . 115 .00634 32 57.40 4.610 . 112 .00481 64 113. 00 8.260 . 110 .00405 96 169.00 11.900 . 110 .00379

128 224.00 15.600 . 110 .00366 160 280.00 19.200 . 109 .00359 192 336.00 22.800 . 109 .00354 224 ·391. 00 26.500 . 109 .00350

CT = 1.0 4 8.70 1. 180 . 136 .01290 8 15.70 1.560 . 122 .00783

16 29.60 2.320 . 115 .00528 32 57.40 3.840 . 112 .00401 64 113.00 6.880 . 110 . 00337 96 169.00 9.920 . 110 .00316

128 224.00 13.000 . 110 .00305 160 280.00 16.000 . 109 .00299 192 336. 00 19.000 . 109 .00295 224 391. 00 22. 100 . 109 .00292

CT = 0.9 4 8.70 1.060 . 136 .01160 8 8.70 1 .060 . 136 .01160

16 29.60 2.090 . 115 .00475 32 57.40 3.460 . 112 .00361 64 113.00 6. 190 . 110 .00303 96 169.00 8.930 . 110 .00284

128 224.00 11. 700 . 110 .00275 160 280.00 14.400 . 109 .00269 192 336. 00 17. 100 . 109 .00265 224 391 .00 19.900 . 109 .00262

Table D-2. IBM 3704/3705 Timing Chart


•

Index

A-factor interference 33 address of next CCW 7 addressing, storage 9 assessing program overrun 19

B-factor interference 33 block-multiplex mode I 0 block-multiplexer channel

implementation I 0 late command chaining 28 loading 28 overrun 28

burst mode 9 byte count 7 byte-mode

channel load limit 36 considerations 29 evaluation procedure 36

byte-multiplexer channel device priority 30 implementation 9 loading 29

capabilities, input/ output 21 card read punch, 1442 29 card reader, 2S01 30 card units 26 CCW flags 7 chaining

command 7 data 7 late command 28

channel block-multiplexer 10 byte-multiplexer 9 byte-multiplexer device priority 30

channel commands control 6 read 6 read backward 6 sense 6 transfer-in-channel (TIC) 6 write 6

channel command words 7 channel command words ( CCW) 7 channel control 6 channel data buffer 11 channel evaluation factors

12SS and 1287 SS, 61 1288 and 1419 S6, 62 1442 and 2S01 Bl and 82 S7, 63 2S20 S8, 64 2701 S9, 6S 7770 60, 66

channel functions S channel interface with processor 16 channel load limit, byte-mode 36 channel load sum worksheet procedure 4S channel loading, byte-multiplexer 29 channel priority 11 channel programs, conventions for satisfactory 21 channel registers 7 channel shared UCW 13 channel status 7 channel trap requests 11 channel UCW directory 13 channel-available channel interruption l 0

channel-control information 6 channel/processor interference percentage 18 channel/ processor interference time 16 channels, evaluating heavily loaded 27 class A commands 23 class B commands 23 class C commands 23 class D commands 23 classes of commands 23 clear 1/0 6 command chaining

late 9 late, block multiplexer 27

command, channel (see channel commands) command classification for 1/0 devices 23 command retry 9 command-chained channel programs l 0 commands

card units 26 class A 23 class B 23 class C 2S class D 2S classes of 23 communication adapters 26 printers 26 tape units 2S

communication adapters 26 concurrent input/ output capabilities 21 control units

2821 12 2848 30 3272 12

conventions for satisfactory channel programs 21

data address 7 data buffer, channel 11 data chaining

considerations 22 description 8 in gaps 8 rates 69

data rate by device 67 device

considerations 13 data rate by 67 end indication 8 load 29 priority on byte-multiplexer channel 30 wait time 29

devices, priority 31 devices, waiting 31 direct-access storage devices

considerations 24 2314 8 3330 8

display control, 2848 30 display station, 2660 30 display, 3277 lS

evaluating heavily loaded channels 27 expected tag sequence 10

figures, list of 4 function of traps 11

Index 71

general channel information S

halt device 6 halt 1/0 6 high channel priority procedure 46

IBM 3704/3705 calculations 49 incorrect length indication 9 initial microcode load (IML) 13 input/ output capabilities 21 instructions, processor (see processor instructions) interference

A-factor 33 B-factor 33 from priority devices 31

1/0 devices 9 1/0 devices, command classifications 23 1/0 interruption 9

late command chaining, byte multiplexer 27 late command chaining 9 load sum 36 loss of performance 21

magnetic character reader, 1419 19 mixed-mode, seven-track tape operation 25 mode

burst 9 multiplex 9 selector 10

multiplex mode 9

normal channel priority procedure 46

overrun description 21 evaluation consideration 28 program 19 testing for 28

pool, ucw 12 preliminary worksheet example 41 previous load 35 printer, 1403 12, 29 printer, 1443 12, 30 printer, 3211 12 printers 26 priority

channel 11 device on byte-multiplexer channel 30 devices 31 devices, interference from 31 load formula 33 loads 31 time factors 32

processor channel characteristics processor, channel interface with processor instructions

clear 1/0 6 halt device 6 halt 1/0 6 start 1/0 6 start I/ 0 fast release 6 test channel 6 test 1/0 6

5 16


program overrun 19 program overrun, assessing 19 programs, command-chained channel 10 protection key 7

ranges of wait times 31 read/punch, 2540 12 relationship of conventions and evaluation procedure 22

scope of conventions 22 select-in line 30 select mode bit 13 selector mode 10 shared bit 15 shared subchannel 12 start 1/0 6 start I/ 0 fast release 6 storage addressing 9 storage control, 3830 12 storage requirements, system 13 subchannel

and unit control words (UCWs) 12 description 10 shared 12 unshared 12

select-out line 30 system storage requirements 13

tables, list of 4 tape units 25 test channel 6 test 1/0 6 testing for overrun 28 transfer-in-channel (TIC) 7 trap request inhibit latch 10 trap requests, channel 10 traps, function of 11

ucw assignment 13 assignment chart 14 channel shared 13 description 12 directory, channel 13 pool 12

unit control words (see UCW) unshared subchannel 12

waiting devices 31 worksheet example, preliminary 41 worksheet procedure, channel load sum 45 worst-case loads 21

1255 channel evaluation factors 1287 channel evaluation factors 1288 channel evaluation factors 1403 printer 12, 29 1419

channel evaluation factors magnetic character reader

1442 card read punch 29 channel evaluation factors

1443 printer 12, 30

SS, 61 55, 61 56,62

56,62 19

57,63

•

•

2501 card reader 30 channel evaluation factors

2520 channel evaluation factors 2540 read/punch 12 2660 display station 30 2701 channel evaluation factors 2821 control unit 12, 30

57, 63 58, 64

59, 65

2848 display control 30 3211 printer 12 3272 control unit 12, 15 3277 display 12, 15 3704/3705 calculations 49 3830 storage control 12 7770 channel evaluation factors 60, 66

Index 73

GA24-3780-1

CJ s: ~ w ~

iJ .., 0 n CD

"' "' 0 • .., s: 0 c.. !!!.. G) .., 0 c: 'O

"' ,l\l

Ql ::i c..

I\)

(") :T Ql ::i ::i !!!.. (") :T Ql ..,,, .., Ql n .... CD .., u;· ~. n "' "Tl

<D z ? ~ w 0 0 6

iJ :!. ::i .... CD c.. ::i

c i:n ~ G) )> I\) ~

w ....... 00 --....------ c;:> ..... - ----- - -- -. ---- -- -~------_ _..._,_

®

•

..

IBM 4341 Processor Model Groups 2, 11, and 12 Channel Characteristics Order No. GA24-3780-l

READER'S COMMENT FORM

This manual is part of a library that serves as a reference source for systems analysts, programmers, and operators of IBM systems. This form may be used to communicate your views about this publication. They will be sent to the author's department for whatever review and action, if any, is deemed appropriate. Comments may be written in your own language; use of English is not required.

IBM may use or distribute any of the information you supply in any way it believes appropriate without incurring any obligation whatever. You may, of course, continue to use the information you supply.

Note: Copies of IBM publications are not stocked at the location to which this form is addressed . Please direct any requests/or copies of publications, or for assistance in using your IBM system, to your I BM representative or to the I BM branch office serving your locality .

• Are you able to find information in this publication easily?

• Are the illustrations adequate, clear, and concise?

• Which of the following are included in your occupational responsibilities? System operating D Instructing Programming D Supervising Maintaining D Other

• If other. please explain.

D D D

Please contact your local IBM branch office for address and subscription service changes.

Your comments:

Note: Staples can cause problems with automated mail sorting equipment. Please use pressure sensitive or other gummed tape to seal this form. If you would like a reply, please supply your name and address on the reverse side of this form.

Thank you for your cooperation. No postage stamp necessary if mailed in the U.S.A. (Elsewhere, an IBM office or representative will be happy to forward your comments.)

GA24-3780-1

Reader's Comment Form

Fold and Tape Please Do Not Staple

I I BUSINESS REPLY MAIL FIRST CLASS PERMIT NO. 40 ARMONK, N.Y.

POSTAGE WILL BE PAID BY ADDRESSEE:

International Business Machines Corporation Department K 10 P.O. Box 6 Endicott, New York 13760

Fold and Tape

NO POSTAGE NECESSARY IF MAILED

IN THE UN.ITED STATES

n ~ e "11 0 ii ~ 0 ~

"' r :; ..

--·-·······························································································································································································' Fold Fold

If you would like a reply, please print:

-~------- - - ------ - -- -. ---- --_ _. _ _____ -~_,_

®

Your Name-----------------------------

Company Name ---------------- Department------

Street Address -------------------ary _~~~~~~~~~~~~~~~~~~~

State ------------- Zip Code ------IBM Branch Office serving you --------------------

OJ s:: .i::. w .i::.

-0

0 0 ct>

"' "' 0 .... s:: 0 a. ~

G> .... 0 c "C

"' -1\,)

"Tl

Ci)

z 0

.i::. w 0 0 6

•

•

•

..

•





Note: Copies of I BM publications are not stocked at the location to which this form is addressed . Please direct any requests for copies of publications, or for assistance in using your I BM system, to your I BM representative or to the I BM branch office serving your locality .



• Which of the following are included in your occupational responsibilities? System operating 0 Instructing Programming 0 Supervising Maintaining 0 Other

• 1 f other, please explain.

0 0 0


Your comments:



GA24-3780-1


Fold and Tape

Fold

Please Do Not Staple

I I BUSINESS REPLY MAIL FIRST CLASS PERMIT NO. 40 ARMONK, N.Y.



Fold and Tape


IN THE UNITED STATES

Fold


}~urName -----------------------------~ Company Name ---------------- Department _____ _

Street Address---------------------City _______________________ _

State ·-------------Zip Code ______ _

-~------- IBM Branch Office serving you ---------------------- -- ---- - -- -. ---- - - --------_ _.._,_ ®

co s: _j:>. w _j:>.

" .., 0 (") • (!)

"' "' 0 .., s: 0 .. a.. ~

G> 0 c 'O "' "' Ql ::J a..

"' 0 ::;; Ql ::J ::J ~

0 ::;; Ql .., Ql (") ...

"""""' (!) .., c;;· !:!". (")

"' 11

Ci)

z ~ _j:>. w 0 0 6

::,0 ::J ... (!)

a.. ::i

c c.n )>

G> )>

"' .j:>. ..

w '1 co 9

•





Note: Copies of IBM publications are not stocked at the location to which this form is addressed. Please direct any requests for copies of publications, or for assistance in using your I BM system, to your I BM representative or to the I BM branch office serving your locality.



• Which of the following are included in your occupational responsibilities? System operating D Instructing Programming D Supervising Maintaining 0 Other

• If other. please explain.

D D D


Your comments:



GA24-3780-1


Fold and Tape Please Do Not Staple

1111

BUSINESS REPLY MAIL FIRST CLASS PERMIT NO. 40 ARMONK, N.Y.



Fold and Tape


IN THE UNITED STATES

n ~ ~ "11 ~ a. ;!? 0

" "' r :; ..

••••••••••••.•••••.••••••.•.•••••.••••.••••.••••..••.•.•.....•.....•••••...•..•.....•..•...•.•.•.•..•..•........•••.•.....•....•..•.••..•....•.•••.••••.•••.••••......•...••..•..••..•..•.•••••••• .1 Fold Fold


--...--------- -- -. ..... ---- -.. ---- - - --------_ _.._,_ ®

Your Name----------------------------~ Company Name---------------- Department _____ _

Street Address ---------------------CTty ~~~~~~~~~~~~~~~~~~~-

State ------------- Zip Code ------IBM Branch Office serving you -------------------

GA24-3780-1

CJ s: """ w

""" -0

0 n ,. CD

"' "' 0 .., ,.. s: 0 a. ~ (;)

0 c:

"O

"' I'.)

..... -Ql

::i a.

I'.)

() :::; Ql

::i ::i ~ () :::; Ql .., Ql

n .... ~.'11111111' "' !:!". n "' "Tl

co 2 ?

""" w 0 0 6

~ ::i r+ CD a. ::i

c i:n ?> (;) )> I'.)

""" ~

w -..J 00

l <? .....

•

IBM 4341 Processor Model Groups 2, 11 and 12 Channel...

Documents

Transcript of IBM 4341 Processor Model Groups 2, 11 and 12 Channel...